A History of Modern Computing.Ceruzzi2003.2e

现代计算史

A History of Modern Computing

计算史

I. Bernard Cohen 和 William Aspray 编辑

History of Computing

I. Bernard Cohen and William Aspray, editors

威廉·阿斯普雷、约翰·冯·诺依曼和现代计算的起源

William Aspray, John von Neumann and the Origins of Modern Computing

Charles J. Bashe、Lyle R. Johnson、John H. Palmer 和 Emerson W. Pugh，IBM 的早期计算机

Charles J. Bashe, Lyle R. Johnson, John H. Palmer, and Emerson W. Pugh, IBM’s Early Computers

Martin Campbell-Kelly，《软件行业历史：从航空预订到刺猬索尼克》

Martin Campbell-Kelly, A History of the Software Industry: From Airline Reservations to Sonic the Hedgehog®

Paul E. Ceruzzi，《现代计算史》

Paul E. Ceruzzi, A History of Modern Computing

I. Bernard Cohen、Howard Aiken：计算机先驱者的肖像

I. Bernard Cohen, Howard Aiken: Portrait of a Computer Pioneer

I. Bernard Cohen 和 Gregory W. Welch 编辑，《创造数字：霍华德·艾肯和计算机》

I. Bernard Cohen and Gregory W. Welch, editors, Makin’ Numbers: Howard Aiken and the Computer

约翰·亨德利，《为失败而创新：政府政策和早期英国计算机产业》

John Hendry, Innovating for Failure: Government Policy and the Early British Computer Industry

迈克尔·林德格伦，《荣耀与失败：约翰·穆勒、查尔斯·巴贝奇、乔治和爱德华·舒茨的差分机》

Michael Lindgren, Glory and Failure: The Difference Engines of Johann Müller, Charles Babbage, and Georg and Edvard Scheutz

David E. Lundstrom，《来自 Univac 的几个好人》

David E. Lundstrom, A Few Good Men from Univac

R. Moreau，《计算机的成熟：人、硬件和软件》

R. Moreau, The Computer Comes of Age: The People, the Hardware, and the Software

Emerson W. Pugh，《打造 IBM：塑造行业及其技术》

Emerson W. Pugh, Building IBM: Shaping an Industry and Its Technology

Emerson W. Pugh，塑造一个行业的记忆

Emerson W. Pugh, Memories That Shaped an Industry

Emerson W. Pugh、Lyle R. Johnson 和 John H. Palmer，IBM 360 和早期 370 系统

Emerson W. Pugh, Lyle R. Johnson, and John H. Palmer, IBM’s 360 and Early 370 Systems

Kent C. Redmond 和 Thomas M. Smith，《从 Whirlwind 到 MITRE：SAGE 防空计算机的研发故事》

Kent C. Redmond and Thomas M. Smith, From Whirlwind to MITRE: The R&D Story of the SAGE Air Defense Computer

Raúl Rojas 和 Ulf Hashagen，《第一台计算机——历史和架构》编辑

Raúl Rojas and Ulf Hashagen, editors, The First Computers—History and Architectures

多萝西·斯坦因，《艾达：一生与遗产》

Dorothy Stein, Ada: A Life and a Legacy

John Vardalas，《加拿大的计算机革命：建设国家技术能力，1945-1980》

John Vardalas, The Computer Revolution in Canada: Building National Technological Competence, 1945–1980

莫里斯·V·威尔克斯，《计算机先驱的回忆录》

Maurice V. Wilkes, Memoirs of a Computer Pioneer

现代计算史

A History of Modern Computing

第二版

Second edition

保罗·E·塞鲁齐

Paul E. Ceruzzi

麻省理工学院出版社

The MIT Press

马萨诸塞州剑桥

Cambridge, Massachusetts

伦敦，英国

London, England

All rights reserved. No part of this book may be reproduced in any form or by any electronic or mechanical means (including photocopying, recording, or information storage and retrieval) without permission in writing from the publisher.

本书由英国索尔兹伯里的 Techset Composition Ltd. 以新巴斯克维尔为背景出版，并在美国印刷和装订。

This book was set in New Baskerville by Techset Composition Ltd., Salisbury, UK, and was printed and bound in the United States of America.

美国国会图书馆出版品目錄數據

Library of Congress Cataloging-in-Publication Data

塞鲁齐（Ceruzzi），保罗 E.

Ceruzzi, Paul E.

现代计算史 / Paul E. Ceruzzi.—第二版

A history of modern computing / Paul E. Ceruzzi.—2nd ed.

页厘米

p. cm.

包括参考书目和索引。

Includes bibliographical references and index.

ISBN 0-262-53203-4 (平装：碱性纸)

ISBN 0-262-53203-4 (pbk. : alk. paper)

1. 计算机—历史。2. 电子数据处理—历史。I. 标题。

1. Computers—History. 2. Electronic data processing—History. I. Title.

QA76.17 .C47 2003

004'.09'049—dc21

QA76.17 .C47 2003

004’.09’049—dc21

2002040799

10 9 8 7 6 5 4 3 2 1

奉献

Dedication

我在史密森尼学会国家航空航天博物馆的办公室里写了这本书，这是世界上最繁忙的公共场所之一。在一个典型的夏日，博物馆可能会有超过 50,000 名游客——相当于一个小城市的人口。这些渴望了解现代技术的游客给了我很大的启发。他们的存在不断提醒我，技术不仅关乎机器，还关乎人：设计和制造机器的人，更重要的是，生活深受机器影响的人。我谨将这本书献给这些游客。

I wrote this book in an office at the Smithsonian Institution’s National Air and Space Museum, one of the busiest public spaces in the world. On a typical summer day there may be upwards of 50,000 visitors to the museum—the population of a small city. These visitors—with their desire to know something of modern technology—were a great inspiration to me. Their presence was a constant reminder that technology is not just about machines but about people: the people who design and build machines and, more importantly, the people whose lives are profoundly affected by them. It is to these visitors that I respectfully dedicate this book.

内容

Contents

奉献精神

Dedication v

第二版前言 ix

Preface to the Second Edition ix

致谢 xiii

Acknowledgments xiii

简介：定义“计算机” 1

Introduction: Defining “Computer” 1

商业计算的出现，1945-1956 13

The Advent of Commercial Computing, 1945–1956 13

计算技术成熟，1956-1964 47

Computing Comes of Age, 1956–1964 47

软件的早期历史，1952–1968 79

The Early History of Software, 1952–1968 79

从大型机到小型机，1959–1969 109

From Mainframe to Minicomputer, 1959–1969 109

“Go-Go”年代和体制/360, 1961–1975 143

The “Go-Go” Years and the System/360, 1961–1975 143

芯片及其影响，1965-1975 177

The Chip and Its Impact, 1965–1975 177

个人电脑，1972–1977 207

The Personal Computer, 1972–1977 207

增强人类智力，1975-1985 243

Augmenting Human Intellect, 1975–1985 243

工作站、UNIX 和网络，1981 年至 1995 年 281

Workstations, UNIX, and the Net, 1981–1995 281

“互联网时代”，1995-2001 307

“Internet Time,” 1995–2001 307

结论：世界图景的数字化 345

Conclusion: The Digitization of the World Picture 345

注释 351

Notes 351

参考书目 415

Bibliography 415

索引 431

Index 431

第二版前言

Preface to the Second Edition

当我完成《现代计算史》第一版的手稿时，我发现自己焦急地回头张望，担心计算领域的某些新发展会使我刚刚写的内容过时。我的担心是有道理的：在我写最后一章的时候，至少发生了一件事，威胁到我建立的叙事结构。那就是微软在 1997 年秋天推出 4.0 版 Internet Explorer 时引发的轰动——这一发布导致美国司法部对该公司提起反垄断诉讼。当时我并没有太关注微软的 Web 战略，但就在我将完成的《现代计算史》手稿放入联邦快递包裹寄给出版商的那一天，我面对着围绕 Internet Explorer 的兴奋。事实上，反垄断诉讼确实是 1995 年以来计算领域最大的发展之一，本期将对此进行详细分析。现在是否还有其他发展潜伏在幕后，而当它们浮现出来时，将使得任何编写计算历史的尝试都变得不可能？

As I was completing the manuscript for the first edition of A History of Modern Computing, I found myself anxiously looking over my shoulder, worrying that some new development in computing would render what I had just written obsolete. My concern was well grounded: as I was writing the final chapter, at least one event occurred that threatened to upset the narrative structure I had erected. That was the fanfare that surrounded Microsoft’s introduction, in the fall of 1997, of version 4.0 of its Internet Explorer—an introduction that led the U.S. Justice Department to file an antitrust suit against the company. I had not been paying much attention to Microsoft’s Web strategy at the time, but I was confronted with the excitement surrounding Internet Explorer literally on the day I put my completed manuscript of A History of Modern Computing into a FedEx package for shipment to the publisher. The antitrust suit did in fact turn out to be one of the biggest developments in computing since 1995, and this edition will examine it at length. Are other developments now lurking in the background, which, when they surface, will render any attempt to write a history of computing impossible?

随着万维网的兴起，出现了“互联网时间”的概念。Netscape 的创始人 Jim Clark 在他 1999 年出版的同名书中将其称为“Netscape 时间”：他将其定义为一项技术从发明到原型、生产、商业成功、成熟和衰落的时间的缩减。1^历史学家面临着芝诺悖论的现代版本。在经典故事中，一个快跑者永远无法到达比赛的终点线，因为他必须先跑完一半的距离，这需要有限的时间，然后再跑完剩余距离的一半，这又需要更短但仍然是有限的时间，依此类推。从将完成的手稿发送给排字员到将书籍或期刊文章交付给读者之间有一段有限的时间。当主题是计算时，芝诺悖论就会占据主导地位：在那短暂的时间间隔内发生的足够多的事情使刚刚写的内容过时。许多人认识到了这一点，并接受了电子出版的解决方案，从而将时间压缩到零。确实有许多网站致力于计算机历史，其中一些质量非常好。然而，接受网络出版是一种虚假的希望，因为它并不能压缩将历史材料组织成连贯叙述所花的时间。历史是事实的编年史，但历史这个词包含了故事这个词，讲故事不会因技术而过时。讲故事的人既不能也不应该加快这一活动的速度。

With the rise of the World Wide Web came the notion of ‘‘Internet Time.’’ Netscape’s founder Jim Clark called it ‘‘Netscape Time’’ in his 1999 book by that title: he defined it as a telescoping of the time for a technology to proceed from invention to prototype, production, commercial success, maturity, and senescence.¹ The historian faces a modern version of Zeno’s paradox. In the classical story, a fast runner never reached the finish line in a race, because he first had to traverse one-half the distance to the end, which took a finite time, and then one-half the remaining distance, which again took a smaller but still finite time, and so on. There is a finite time between sending a completed manuscript to the typesetter and the delivery of a book or journal article to the reader. When the subject is computing, Zeno’s paradox takes control: enough happens in that brief interval to render what was just written obsolete. Many recognize this and embrace the solution of publishing electronically, thus telescoping that time down to zero. There are indeed many Web sites devoted to the history of computing, some of excellent quality. Still, embracing Web publishing is a false hope, because it does nothing to compress the time spent organizing historical material into a coherent narrative. History is a chronology of facts, but the word history contains the word story in it, and telling stories is not rendered obsolete by technology. The storyteller neither can, nor should, speed that activity up.

回顾第一版，我觉得它成功避开了芝诺的陷阱。1995 年之后发生了许多重大事件，在新的一章中，我将详细探讨其中三件。它们是上面提到的微软审判案；“dot.com”公司的爆炸式增长和同样令人震惊的崩溃；以及“开源”软件运动的兴起，尤其是 Linux 操作系统的采用。这些是我可以选择的至少十几个主题中的三个，但探讨更多内容对读者没有好处。

In looking over the first edition, I feel that it has managed to avoid Zeno’s trap. A number of significant events have developed after 1995, and in a new chapter I examine three at length. These are the Microsoft trial, mentioned above; the explosion and equally stunning implosion of the ‘‘dot.com’’ companies; and the rise of the ‘‘open source’’ software movement and especially the adoption of the Linux operating system. These are three of at least a dozen topics that I could have chosen, but to examine more would not serve the reader.

芝诺也许会报仇雪恨。上述更新计算机历史的计划看似合理，但可能存在致命缺陷。作为一门独立学科，计算机历史本身可能变得无关紧要。有充分的证据证明这一点。例如，当财经媒体提到“科技”股票时，它不再指 IBM 甚至英特尔等公司所代表的计算机行业，而是越来越多地指互联网和电信公司。在我担任博物馆馆长的工作中，我不得不努力解决如何利用文物向公众展示计算机历史的问题。当问题在于计算机是矩形的“黑匣子”时，这已经够困难了，因为计算机的功能很少；现在，这个故事似乎全是关于“网络空间”的，而从定义上讲，它没有有形的性质。

Zeno may get his revenge yet. The above plan for bringing the history of computing up to date seems rational, but it may have a fatal flaw. The history of computing, as a separate subject, may itself become irrelevant. There is no shortage of evidence to suggest this. For example, when the financial press refers to ‘‘technology’’ stocks, it no longer means the computer industry represented by companies like IBM or even Intel, but increasingly Internet and telecommunications firms. In my work as a museum curator, I have had to grapple with issues of how to present the story of computing, using artifacts, to a public. It was hard enough when the problem was that computers were rectangular ‘‘black boxes’’ that revealed little of their function; now the story seems to be all about “cyberspace,” which by definition has no tangible nature to it.

也许计算机的发明就像尼古拉斯·奥托于 1876 年发明的四冲程汽油发动机。无论这有多重要，如果奥托被人们记住，那是因为奥托循环成为汽车的首选动力方式。而汽车也是一个值得研究的课题，不仅因为它作为机器的内在品质，还因为它有助于塑造一个围绕个人交通工具发展起来的社会。在第一版的前言中，我曾提出，本书强调从批处理到交互式计算的转变，有朝一日可能会成为计算历史的次要部分。这一天已经到来了吗？现在看来至关重要的是计算机从独立设备到联网设备的转变。然而，如果没有早期从批处理到交互式使用的转变，这一转变就不可能发生。虽然网络空间的硬件根源在于英特尔等芯片制造商和苹果等个人电脑公司，但网络空间的精神根源在于 MAC 项目等分时实验。

Perhaps the invention of the computer is like Nicholaus Otto’s invention of the four-cycle gasoline engine in 1876. However significant that was, if Otto is remembered at all it is because the Otto Cycle became the preferred way to power the automobile. And the automobile in turn is a topic worthy of study not so much for its intrinsic qualities as a machine, but for helping shape a society that has grown around personal transportation. In the preface to the first edition I suggested that this book’s emphasis on the transition from batch-oriented to interactive computing might some day seem to be a minor part of computing history. Has that day come already? What now seems to have been critical was the transformation of the computer from a stand-alone to a networked device. That, however, could not have happened were it not for the earlier transition from batch to interactive use. Although the hardware roots of cyberspace are found in chip manufacturers including Intel, and in personal computer companies like Apple, the spiritual roots of cyberspace are found in time-sharing experiments like Project MAC.

我不认为计算机历史会消失在网络空间历史的一个分支中。最近网络公司的倒闭（新章节中涉及的第二个主题）表明，对硬件和软件（包括 Linux，第三个主题）的研究仍将处于任何历史的核心地位。随着数字技术越来越多地介入人类互动，网络空间研究正在与社会、文化、军事和政治历史融合。这就是“媒体”一词的起源。我希望这本书将继续服务于那些想知道我们现在生活的日益媒介化的世界是如何产生的人们。

I do not feel that the history of computing will vanish into a subfield of the history of cyberspace. The recent implosion of the dot.com companies (the second topic covered in the new chapter) suggests that a study of hardware and software (including Linux, the third topic) will remain at the core of any history. The study of cyberspace is merging with social, cultural, military, and political history, as digital technologies increasingly mediate among human interactions. That is the origin of the term media. I hope this book will continue to serve those who wish to know how the increasingly mediated world we now live in arose.

致谢

Acknowledgments

许多机构和人士协助我撰写了本书。

Many institutions and people assisted me in the writing of this book.

我得到了国家档案馆、国会图书馆手稿部、艾森豪威尔图书馆、哈格利博物馆和图书馆、查尔斯·巴贝奇研究所、计算机博物馆、加利福尼亚州圣何塞公共图书馆、斯坦福大学档案馆、美国国家历史博物馆档案馆、数字设备公司档案馆和法国图书馆的帮助，找到了手稿和文件。

I received help in locating manuscripts and papers from the National Archives, the Library of Congress Manuscript Division, the Eisenhower Library, the Hagley Museum and Library, the Charles Babbage Institute, The Computer Museum, the San Jose, California, Public Library, the Stanford University Archives, the National Museum of American History Archives, Digital Equipment Corporation Archives, and the French Library.

我收到了来自 Tim Bergin、Robert Smith、Bruce Seely、Michael Neufeld、Allan Needell、Jon Eklund、Steve Lubar、《科技与文化》和《历史与科技》的匿名审稿人、美国国家航空航天博物馆当代历史研讨会成员、Bill Aspray、Martin Campbell-Kelly、JAN Lee、Eric Weiss、Gordon and Gwen Bell、Herb Grosch、Mike Williams、Paul Forman、Oscar Blumtritt、John Wharton 和 GP Zachary 等对章节草稿或整个手稿的有益评论。

I received helpful comments on drafts of chapters or the whole manuscript from Tim Bergin, Robert Smith, Bruce Seely, Michael Neufeld, Allan Needell, Jon Eklund, Steve Lubar, anonymous referees at Technology and Culture and at History and Technology, members of the National Air and Space Museum’s Contemporary History Seminar, Bill Aspray, Martin Campbell-Kelly, J. A. N. Lee, Eric Weiss, Gordon and Gwen Bell, Herb Grosch, Mike Williams, Paul Forman, Oscar Blumtritt, John Wharton, and G. P. Zachary.

哈罗德·莫丁 (Harold Motin)、托尼·托马斯 (Toni Thomas) 以及国家航空航天博物馆摄影服务部的工作人员为我的照片研究提供了帮助。

I was assisted in my photo research by Harold Motin, Toni Thomas, and the staff of the National Air and Space Museum’s Photographic Services Department.

我还要感谢以下人士的支持：Diane Wendt、Peggy Kidwell、Connie Carter、Michael Mahoney、Alice Jones 和 Jamie Parker Pearson。我还要感谢麻省理工学院出版社的 Larry Cohen、Sandra Minkkinen 和 Chryseis O. Fox 为这个项目付出的辛勤工作和奉献。

I also wish to acknowledge the following persons for their support: Diane Wendt, Peggy Kidwell, Connie Carter, Michael Mahoney, Alice Jones, and Jamie Parker Pearson. I wish also to thank Larry Cohen, Sandra Minkkinen, and Chryseis O. Fox, of the MIT Press, for their hard work and dedication to this project.

部分资金支持来自史密森尼学会董事会出版基金。

Financial Support came in part from the Smithsonian Institution’s Regents’ Publication Fund.

简介：定义“计算机”

Introduction: Defining “Computer”

计算机被发明来“计算”：解决“复杂的数学问题”，正如字典中对这个词的定义。1^它们仍然在做这件事，但这并不是我们生活在“信息时代”的原因。这反映了计算机所做的其他事情：存储和检索数据、管理通信网络、处理文本、生成和处理图像和声音、驾驶飞机和航天器等等。计算机内部的电路通过将这些事情转换成数学语言来完成这些工作。但我们大多数人从未见过这些方程式，即使我们见过，也很少有人能理解它们。然而，我们大多数人都参与了这种数字文化，无论是使用 ATM 卡、编写和打印办公室通讯、拨打免费电话给邮购公司订购第二天送达的衣服，还是在库存“及时”补货的大型商场购物。对于这些和许多其他应用，我们可以充分利用这项发明的所有功能，而无需看到方程式。就公众形象而言，“计算”是计算机所做的最不重要的事情。

Computers were invented to “compute”: to solve “complex mathematical problems,” as the dictionary still defines that word.¹ They still do that, but that is not why we are living in an “Information Age.” That reflects other things that computers do: store and retrieve data, manage networks of communications, process text, generate and manipulate images and sounds, fly air and space craft, and so on. Deep inside a computer are circuits that do those things by transforming them into a mathematical language. But most of us never see the equations, and few of us would understand them if we did. Most of us, nevertheless, participate in this digital culture, whether by using an ATM card, composing and printing an office newsletter, calling a mail-order house on a toll-free number and ordering some clothes for next-day delivery, or shopping at a mega-mall where the inventory is replenished “just-in-time.” For these and many other applications, we can use all the power of this invention without ever seeing an equation. As far as the public face is concerned, “computing” is the least important thing that computers do.

但电子数字计算机是为了解方程而发明的。“计算机”一词最初是指解方程的人；直到 1945 年左右，这个名字才被沿用到机器上^。2

But it was to solve equations that the electronic digital computer was invented. The word “computer” originally meant a person who solved equations; it was only around 1945 that the name was carried over to machinery.²

一项发明在社会上获得了其发明者未曾预见到的地位，这并不奇怪。3^计算机的故事就说明了这一点。并不是说计算机最终没有用于计算——今天大多数从事计算工作的科学家和工程师都在使用它。至少计算机的发明者预测到了这一点。但人们找到了让这项发明发挥更多作用的方法。他们如何做到这一点，将 20 世纪 40 年代的数学引擎转变为 20 世纪 90 年代的网络信息设备，正是本书的主题。

That an invention should find a place in society unforeseen by its inventors is not surprising.³ The story of the computer illustrates that. It is not that the computer ended up not being used for calculation—it is used for calculation by most practicing scientists and engineers today. That much, at least, the computer’s inventors predicted. But people found ways to get the invention to do a lot more. How they did that, transforming the mathematical engines of the 1940s to the networked information appliance of the 1990s, is the subject of this book.

图 0.1

20 世纪 50 年代早期，洛杉矶北美航空公司的“人工计算机”正在工作。照片下方中央的两个女人正在使用 Friden 计算器，左下方的男人正在查找计算尺上的数字。房间的后半部分用于绘图。这个房间里没有任何打孔卡机，尽管航空工程师确实将它们用于一些应用，如文中所述。（来源：NASM。）

Figure 0.1

Human “computers” at work at North American Aviation, Los Angeles, in the early 1950s. The two women in the lower center of the photo are using Friden calculators, and the man at the lower left is looking up a number on a slide rule. The rear half of the room is set up for drafting. Absent from this room are any punched-card machines, although aircraft engineers did use them for some applications, as described in the text. (Source : NASM.)

计算机革命与技术史

The Computer Revolution and the History of Technology

20 世纪 80 年代初，我刚在州立大学历史系找到第一份工作，当时我跟一位同事提到，我正在研究计算机历史。“为什么要研究计算机？”他回答道，“为什么不研究洗衣机的历史呢？”我以为他在开玩笑，也许是在取笑刚进教职员休息室的菜鸟。但他是认真的。毕竟，他家里有一台洗衣机，它是一件复杂的技术产品，对他的日常生活影响深远。它肯定有历史。但计算机呢？那些是他听说过但只是间接体验过的奇特事物。

In the early 1980s, when I had just taken my first job as a member of the history department of a state university, I mentioned to one of my colleagues that I was studying the history of computing. “Why computing?” he replied. “Why not study the history of washing machines?” I thought he was joking, maybe making fun of the greenhorn just arrived in the faculty lounge. But he was serious. After all, he had a washing machine at home, it was a complex piece of technology, and its effect on his daily life was profound. Surely it had a history. But computers? Those were exotic things he had heard of but experienced only indirectly.

在 20 世纪 90 年代，人们不会问这个问题，因为很少有人会认为计算机不重要。我们生活在一个被计算机改变的时代。4^这也是我们需要了解计算机起源的原因。但我并不喜欢“信息时代”或“计算机革命”等术语。它们既误导人，又提供信息。技术革命确实会发生，尽管并不常见。新技术如何在社会中找到一席之地的故事总是比“X 革命”或“X 时代”这些短语所暗示的更为微妙和复杂，其中“X”代表喷气式飞机、核能、汽车、计算机、信息、空间、互联网、微电子等。5^每日新闻倾向于夸大每一种新芯片型号、每一种新软件、每一种网络新进展、计算和娱乐公司之间的每一次联盟可能产生的影响：它们肯定会让我们的生活变得更好。几周后，这些光彩夺目的报道的主题就被遗忘了，取而代之的是一些我们确信是真正转折点的新发展^。6

In the 1990s that question would not be asked, because few would argue that computers are not important. We live in an age transformed by computing.⁴ This is also the reason why we need to understand its origins. But terms like “Information Age” or “Computer Revolution” are not ones I like. They mislead as much as inform. Technological revolutions certainly do occur, though not often. The story of how a new technology finds its place in a society is always more subtle and complex than implied by the phrase “X Revolution,” or “X Age,” where “X” stands for jet aircraft, nuclear energy, automobiles, computers, information, space, the Internet, microelectronics, and so on.⁵ The daily press tends to overstate the possible effects of each new model of a chip, each new piece of software, each new advance in networking, each alliance between computing and entertainment firms: surely they will change our lives for the better. A few weeks later the subject of these glowing reports is forgotten, replaced by some new development that, we are assured, is the real turning point.⁶

然而，谁会否认计算机技术是革命性的呢？简单地衡量一下现代机器的计算能力，就会发现其发展速度是古代或现代其他技术所无法比拟的。美国家庭和办公室安装的计算机数量也显示出类似的增长速度，而且没有放缓的迹象。现代商业航空旅行、税收、医疗管理和研究、军事规划和行动——这些以及许多其他活动都带有计算机支持的烙印，如果没有计算机的支持，它们要么看起来完全不同，要么根本无法执行。计算的历史比洗衣机的历史更能引起公众的关注——也许应该如此。我怀疑，1981 年放弃计算研究的同事现在不会再用手动打字机准备论文了。历史学家是最热衷于接受计算机辅助学术研究最新进展的人之一^。7

Yet who would deny that computing technology has been anything short of revolutionary? A simple measure of the computing abilities of modern machines reveals a rate of advance not matched by other technologies, ancient or modern. The number of computers installed in homes and offices in the United States shows a similar rate of growth, and it is not slowing down. Modern commercial air travel, tax collection, medical administration and research, military planning and operations—these and a host of other activities bear the stamp of computer support, without which they would either look quite different or not be performed at all. The history of computing commands—as it probably should—more attention from the public than the history of the washing machine. The colleague who in 1981 dismissed the study of computing no longer prepares his papers on a manual typewriter, I suspect. Historians are among the most fanatic in embracing the latest advances in computer-based aids to scholarship.⁷

电子计算机是否只是塑造了二十世纪的众多大型高科技系统之一？它作为信息处理机器在多大程度上是独一无二的？1945 年以后的计算与早期的信息处理活动在多大程度上不同？通俗文学往往强调计算的独特性，同时对其革命性影响进行激动人心的描述。一些作家将这场革命描述为“清洁”技术的接管，没有铁器时代技术的任何污染或其他副作用。8如果计算机正在彻底改变我们的生活，^那么谁是失败者；谁是计算机必须从这个新世界中驱逐的忠诚者？或者计算机就像墨西哥的执政党：一场永久的、良性的、制度化的“革命”？我希望接下来的叙述将提供足够的历史数据来回答这些问题，至少是初步回答^。9

Is the electronic computer only one of many large-scale, high-technology systems that have shaped the twentieth century? To what extent is it unique as an information-processing machine? To what extent is computing after 1945 different from the information-handling activities of an earlier age? The popular literature tends to stress computing’s uniqueness, hand in hand with breathless accounts of its revolutionary impacts. Some writers cast this revolution as a takeover by a “clean” technology, with none of the pollution or other side effects of the technologies of the Iron Age.⁸ If the computer is revolutionizing our lives, who is on the losing side; who are the loyalists that computing must banish from this new world? Or is computing like the ruling party of Mexico: a permanent, benign, institutionalized “revolution”? The narrative that follows will, I hope, provide enough historical data to answer these questions, at least tentatively.⁹

当前对计算的研究对这些问题给出了相互矛盾的答案。有些研究显示现代计算与之前的信息处理机制和社会环境之间存在许多联系。10^有些研究顺便提到计算是许多源自二战研究的技术之一。许多人强调计算与战时武器实验室的其他产品之间的区别；很少有人研究它们有什么共同之处。11^还有一些人根本不去尝试发现任何联系^。12

Current studies of computing give conflicting answers to these questions. Some show the many connections between modern computing and the information-handling machinery and social environments that preceded it.¹⁰ Some make passing references to computing as one of many technologies that owe their origins to World War II research. Many stress the distinction between computing and other products of wartime weapons laboratories; few examine what they have in common.¹¹ Still others make little attempt to discover any connection at all.¹²

托马斯·帕克·休斯 (Thomas Parke Hughes) 在论述美国和欧洲电力系统的出现时，引入了技术系统的概念，即特定的机器必须适合于该技术系统。13^他的工作内容过于丰富和复杂，无法在此一一总结，但有几个方面与计算历史特别相关。其中之一是“发明家”包括在社会、政治、经济以及技术领域进行创新的人。有时，硬件的发明者也是这些领域的先驱，有时则不是。在计算历史上，特别是在讨论硅谷的崛起时，我们会一次又一次地遇到与技术创新者有着复杂关系的企业家。本文还将借鉴休斯的另一个见解：尽管人们经常用“里程碑”这样的术语来描述技术，但技术的发展是沿着一条宽阔的前沿而不是线性路径前进的。

In writing about the emergence of electrical power systems in the United States and Europe, Thomas Parke Hughes introduced the notion of technological systems, into which specific pieces of machinery must fit.¹³ His work is too rich and complex to be summarized here, but a few aspects are particularly relevant to the history of computing. One is that “inventors” include people who innovate in social, political, and economic, as well as in technical, arenas. Sometimes the inventor of a piece of hardware is also the pioneer in these other arenas, and sometimes not. Again and again in the history of computing, especially in discussing the rise of Silicon Valley, we shall encounter an entrepreneur with a complex relationship to a technical innovator. This narrative will also draw on another of Hughes’s insights: that technology advances along a broad front, not along a linear path, in spite of terms like “milestone” that are often used to describe it.

然而，在这种系统方法下，计算历史存在问题。现代计算机的一个定义是它是一个系统：硬件和软件在层次结构中的排列。那些在一个层次上使用系统的人看不到或不关心其他层次上发生的事情。最高层由“软件”组成——根据定义，软件没有有形形式，但最好被描述为组织方法。因此，有人可能会说，人们不需要付出任何特殊的努力来将系统方法应用于计算历史，因为系统会自然地出现在任何地方。这是计算独特性的另一个例子。然而，系统方法将应用于本叙述，因为它有助于我们摆脱将计算仅仅视为纯技术领域发明家的产物的观点。

The history of computing presents problems under this systems approach, however. One definition of a modern computer is that it is a system: an arrangement of hardware and software in hierarchical layers. Those who work with the system at one level do not see or care about what is happening at other levels. The highest levels are made up of “software”—by definition things that have no tangible form but are best described as methods of organization. Therefore, it might be argued, one need not make any special effort to apply the systems approach to the history of computing, since systems will naturally appear everywhere. This is another example of computing’s uniqueness. Nevertheless, the systems approach will be applied in this narrative, because it helps us get away from the view of computing solely as a product of inventors working in a purely technical arena.

另一种技术史研究方法被称为“社会建构”。与系统方法一样，它的内容非常丰富，无法在此一一概括。14^简而言之，社会建构主义方法强调，设计计算机系统或将其集成到社交网络中没有“最佳”方法。稳定的配置（例如，当前桌面系统及其软件的使用）既是各种群体（包括工程师）之间社会和政治协商的结果，也是最有效或技术上最佳设计的自然出现。少数计算机历史学家采用了这种方法，¹⁵但大多数没有，他们更喜欢将计算机历史描述为一系列技术问题，这些问题由事后看来自然而然的工程解决方案解决。

Another approach to the history of technology is known as “social construction.” Like the systems approach, it is too rich a subject to be summarized here.¹⁴ Briefly a social constructionist approach to the history of computing would emphasize that there is no “best” way to design computing systems or to integrate them into social networks. What emerges as a stable configuration—say, the current use of desktop systems and their software—is as much the result of social and political negotiation among a variety of groups (including engineers) as it is the natural emergence of the most efficient or technically best design. A few historians of computing have adopted this approach,¹⁵ but most have not, preferring to describe computing’s history as a series of technical problems met by engineering solutions that in hindsight seem natural and obvious.

然而，围绕较近的计算机历史发展起来的大量历史文献确实采用了社会建构主义方法，尽管只是非正式的。个人计算机的出现一直是畅销书和文章的主题，这些作者要么不熟悉关于社会建构的学术辩论，要么知道但避免向外行观众介绍该理论。他们关于个人电脑的故事强调了年轻人的理想主义抱负，这些年轻人主要集中在旧金山湾区，并充满了 20 世纪 60 年代末伯克利言论自由运动的价值观。对于这些作家来说，个人电脑与其说是来自工程师的工作台，不如说是来自 1975 年至 1977 年期间的自制计算机俱乐部会议。16^这些历史往往忽略了固态电子等领域的进步，在这些领域，技术问题以及不同的社会力量发挥了重要作用。他们也没有将美国国防部和美国国家航空航天局（硅谷最大的两个雇主）在塑造技术方面的作用纳入其中。这些联邦机构代表的是社会和政治驱动因素，而非工程驱动因素。我将在下文中借鉴休斯的社会建构概念及其系统方法；我们将发现社会力量发挥作用的大量证据，不仅在个人电脑时代，而且在个人电脑时代之前和之后也是如此。

However, a body of historical literature that has grown around the more recent history of computing does adopt a social constructionist approach, if only informally. The emergence of personal computing has been the subject of popular books and articles by writers who are either unfamiliar with academic debates about social construction or who know of it but avoid presenting the theory to a lay audience. Their stories of the personal computer emphasize the idealistic aspirations of young people, mainly centered in the San Francisco Bay area and imbued with the values of the Berkeley Free Speech Movement of the late 1960s. For these writers, the personal computer came not so much from the engineer’s workbench as from sessions of the Homebrew Computer Club between 1975 and 1977.¹⁶ These histories tend to ignore advances in fields such as solid state electronics, where technical matters, along with a different set of social forces, played a significant role. They also do little to incorporate the role of the U.S. Defense Department and NASA (two of the largest employers in Silicon Valley) in shaping the technology. These federal agencies represent social and political, not engineering, drivers. I shall draw on Hughes’s concepts of social construction and his systems approach throughout the following narrative; and we will find abundant evidence of social forces at work, not only during the era of personal computing but before and after it as well.

主题

Themes

接下来的叙述是按时间顺序进行的，从 20 世纪 40 年代末首次尝试将电子计算机商业化开始，到 20 世纪 90 年代中期联网个人工作站普及为止。我已经确定了几个主要的转折点，这些转折点得到了最密切的关注。它们包括 20 世纪 40 年代末计算机从科学专用仪器转变为商业产品、20 世纪 60 年代末小型系统的出现、20 世纪 70 年代个人计算的出现以及 1985 年以后网络的普及。我还确定了这些变化中一直存在的几个共同线索。

The narrative that follows is chronological, beginning with the first attempts to commercialize the electronic computer in the late 1940s and ending in the mid–1990s, as networked personal workstations became common. I have identified several major turning points, and these get the closest scrutiny. They include the computer’s transformation in the late 1940s from a specialized instrument for science to a commercial product, the emergence of small systems in the late 1960s, the advent of personal computing in the 1970s, and the spread of networking after 1985. I have also identified several common threads that have persisted throughout these changes.

第一条线索与计算机本身的内部设计有关：电子电路的排列方式使机器能够有效可靠地运行。尽管从真空管到集成电路的实现方式发生了变化，但计算机内部的信息流至少在一个层面上没有改变。这种设计被称为“冯·诺依曼架构”，以约翰·冯·诺依曼（1903-1957）的名字命名，他在 1945 年和 1946 年撰写的一系列报告中阐述了这种架构。17^它在底层硬件和软件的连续变化浪潮中仍然存在，为历史学家提供了至少一条通往近代历史茂密森林的道路。一代又一代的机器如何背离 1945 年的概念，同时又保留了它们的本质，这也是故事的主要部分。

The first thread has to do with the internal design of the computer itself: the way that electronic circuits are arranged to produce a machine that operates effectively and reliably. Despite the changes in implementation from vacuum tubes to integrated circuits, the flow of information within a computer, at one level at least, has not changed. This design is known as the “von Neumann Architecture,” after John von Neumann (1903–1957), who articulated it in a series of reports written in 1945 and 1946.¹⁷ Its persistence over successive waves of changes in underlying hardware and software provides the historian with at least one path into the dense forest of recent history. How successive generations of machines departed from the concepts of 1945, while retaining their essence, also forms a major portion of the story.

许多计算机历史都提到了三代计算机，这三代计算机是根据计算机使用的是真空管、晶体管还是集成电路而划分的。事实上，第三代计算机的寿命比前两代计算机的总和还要长，而且似乎还没有任何计算机能够真正挑战硅芯片。硅集成电路封装在矩形黑色塑料封装中，焊接在电路板上，电路板又插入一组称为总线的电线中：这种物理结构自 20 世纪 70 年代以来就已成为标准。然而，它的功能发展迅速，芯片的数据存储容量大约每 18 个月就会翻一番。一些工程师认为，基础硅技术的这种创新速度是历史的真正驱动力，它使计算的新阶段看起来像成熟的果实从树上掉下来一样。这种观点与技术史学家的观点相矛盾，但工程师们接受甚至推广这种观点的程度使其成为一个令人信服的论点，如果不仔细研究，就无法驳斥。

Many histories of computing speak of three “generations,” based on whether a computer used vacuum tubes, transistors, or integrated circuits. In fact, the third of these generations has lasted longer than the first two combined, and nothing seems to be on the horizon that will seriously challenge the silicon chip. Silicon integrated circuits, encased in rectangular black plastic packages, are soldered onto circuit boards, which in turn are plugged into a set of wires called a bus: this physical structure has been a standard since the 1970s. Its capabilities have progressed at a rapid pace, however, with a doubling of the chip’s data storage capacity roughly every eighteen months. Some engineers argue that this pace of innovation in basic silicon technology is the true driving force of history, that it causes new phases of computing to appear like ripe fruit dropping from a tree. This view is at odds with what historians of technology argue, but the degree to which it is accepted and even promoted by engineers makes it a compelling argument that cannot be dismissed without a closer look.

图 0.2

十年后同一家公司的机器计算。一对 IBM 7090 计算机协助设计和测试火箭发动机，这些发动机后来将载人往返月球。此场景中最显眼的物体是磁带驱动器，这表明信息的存储和检索与算术一样都是“计算”的一部分。显然，这间屋子里的人比上一张照片中少，但值得注意的是，这里可见的四个人中，有两个是 IBM 的员工，而不是北美航空公司的员工。（来源：罗伯特·凯利，《生活》杂志，© 时代公司，1962 年。）

Figure 0.2

Computing with machines at the same company ten years later. A pair of IBM 7090 computers assist in the design and testing of the rocket engines that will later take men to the Moon and back. The most visible objects in this scene are the magnetic tape drives, suggesting that storage and retrieval of information are as much a part of “computing” as is arithmetic. Obviously fewer people are visible in this room than in the previous photo, but it is worth noting that of the four men visible here, two are employees of IBM, not North American Aviation. (Source : Robert Kelly, Life Magazine, © Time Inc., 1962.)

1945 年后，美国的计算机技术在繁荣的气氛和强劲的消费市场中发展起来。当时正值与苏联的冷战时期。计算机技术的发展如何适应这种气氛是本文的另一个主题。ENIAC 本身是开启这个时代的机器，它是为了满足军事需求而制造的；紧随其后的是其他对计算机技术产生重大影响的军事项目和武器系统：旋风计划、民兵弹道导弹、高级研究计划局的 ARPANET 等。与此同时，主导计算机技术的 IBM 公司通过专注于商业市场而非军事市场来积累财富和权力，尽管它也从军事合同中获得了可观的收入。20 世纪 70 年代，当军方补贴计算机开发时，美国政府的另一个部门司法部试图拆分 IBM，指控它变得过于庞大和强大。

Computing in the United States developed after 1945 in a climate of prosperity and a strong consumer market. It was also during the Cold War with the Soviet Union. How the evolution of computing fits into that climate is another theme of this story. The ENIAC itself, the machine that began this era, was built to meet a military need; it was followed by other military projects and weapons systems that had a significant impact on computing: Project Whirlwind, the Minuteman ballistic missile, the Advanced Research Projects Agency’s ARPANET, and others. At the same time, the corporation that dominated computing, IBM, built its wealth and power by concentrating on a commercial rather than a military market, although it too derived substantial revenues from military contracts. In the 1970s, as the military was subsidizing computer development, another arm of the U.S. government, the Justice Department, was attempting to break up IBM, charging that it had become too big and powerful.

军方在固态电子技术进步中的作用众所周知，但仔细观察就会发现，这一作用很复杂，而且并不总是有益的。18 “军方”^一词具有误导性：它不是一个单一的军事实体，而是一群经常在角色、任务和资金问题上相互争执的军种和局。由于军事官僚机构庞大而繁琐，能够打破官僚主义的个人“产品冠军”至关重要。海曼·里科弗 (Hyman Rickover) 在为海军研发核动力潜艇方面所起的作用就是一个著名的例子。军事支持也表现为不同的形式。有时强调基础研究，有时强调特定产品。几十年来，这种关系发生了变化，首先是新兴的冷战，然后是朝鲜战争、太空竞赛、越南战争等等。从 1945 年开始，就一直有人认为计算技术可以满足军事需求，因此引导军事资金来推进它是适当的。但并非所有人都认同这一观点，正如以下章节将会展示的那样；但军事支持一直是一个不变的因素。

The military’s role in the advancement of solid state electronics is well known, but a closer look shows that role to be complex and not always beneficial.¹⁸ The term “military” is misleading: there is no single military entity but rather a group of services and bureaus that are often at odds with one another over roles, missions, and funding. Because the military bureaucracy is large and cumbersome, individual “product champions” who can cut through red tape are crucial. Hyman Rickover’s role in developing nuclear-powered submarines for the Navy is a well-known example. Military support also took different forms. At times it emphasized basic research, at others specific products. And that relationship changed over the decades, against a backdrop of, first, the nascent Cold War, then the Korean conflict, the Space Race, the Viet Nam War, and so on. From 1945 onward there have always been people who saw that computing technology could serve military needs, and that it was therefore appropriate to channel military funds to advance it. Not everyone shared that view, as the following chapters will show; but military support has been a constant factor.

苏联解体和冷战结束使这场冲突中一些被隐藏或压制的方面成为焦点。其中之一是科学家和大学研究人员在支持这项工作方面发挥了异常积极的作用。19 另一个是^信息以及信息处理机器所发挥的独特作用。信息或“情报”一直是所有战争的重要组成部分，但只是达到目的的一种手段。在冷战中，它本身就是目的。这是一场密码破译、间谍卫星、模拟和“战争游戏”的战争。基于科学的武器开发和模拟的作用都为美国国防部提供了强大的激励，使其大量投资于数字计算，作为客户，更重要的是，作为基础研究的资金来源^。20

The breakup of the Soviet Union and the end of the Cold War have brought into focus some aspects of that conflict that had been hidden or suppressed. One was the unusually active role played by scientists and university researchers in supporting the effort.¹⁹ Another was the unique role that information, and by implication, information-handling machines, played. Information or “intelligence” has been a crucial part of all warfare, but as a means to an end. In the Cold War it became an end in itself. This was a war of code-breaking, spy satellites, simulations, and “war games.” Both science-based weapons development and the role of simulation provided a strong incentive for the U.S. Defense Department to invest heavily in digital computing, as a customer and, more importantly, as a source of funds for basic research.²⁰

IBM 的角色是本故事中另一个反复出现的主题，它从 1952 年到 20 世纪 80 年代一直主导着计算机行业。它的崛起和 1990 年后出人意料的失败已成为许多书籍的主题。IBM 本身赞助了一系列出色的公司历史，揭示了它如何运作。21^一个问题是，IBM 是一家规模庞大、结构严密、拥有一流研究设施的组织，它如何与威廉·盖茨三世、威廉·诺里斯、肯·奥尔森或马克斯·帕列夫斯基等企业家领导的初创公司抗衡。这些人能够克服 IBM 设置的进入计算机行业的障碍，而其他许多人尝试过但都失败了。除了运气之外，还有什么因素造成了这种差异？IBM 如何在不断且经常颠覆性的技术创新环境中继续占据主导地位？与初创公司不同，IBM 需要担心现有的客户群，这使得它永远无法从头开始。 IBM 的工程和销售团队必须保持与战前支柱打孔卡业务的连续性，即使其自己的研究实验室正在开发新技术，这些技术将取代打孔卡。同样，初创公司一旦建立并取得成功，就会面临同样的处理新技术的问题。因此，我们可以将 IBM 的战略与其新竞争对手的战略进行比较，包括数字设备公司、王安实验室、控制数据和微软。

The role of IBM, which dominated the computer industry from about 1952 through the 1980s, is another recurring theme of this narrative. Its rise and its unexpected stumble after 1990 have been the subject of many books. IBM itself sponsored a series of excellent corporate histories that reveal a great deal about how it operated.²¹ One issue is how IBM, a large and highly structured organization with its own first-class research facilities, fared against start-up companies led by entrepreneurs such as William Gates III, William Norris, Ken Olsen, or Max Palevsky. These people were able to surmount the barriers to entry into the computer business that IBM erected, while a host of others tried and failed. What, besides luck, made the difference? How did IBM continue to dominate in an environment of constant and often disruptive technological innovation? Unlike the start-up companies, IBM had an existing customer base to worry about, which prevented it from ever starting with a clean slate. Its engineering and sales force had to retain continuity with the punched-card business that had been its prewar mainstay, even as its own research laboratories were developing new technologies that would render punched cards obsolete. Likewise the start-up companies, once they became established and successful, faced the same problem of dealing with new technology. We may thus compare IBM’s strategy with the strategies of its new competitors, including Digital Equipment Corporation, Wang Labs, Control Data, and Microsoft.

另一个主题涉及一个在 20 世纪 40 年代末尚不为人所知、但在 20 世纪 90 年代却主导着计算领域的术语：软件。22第 3 章^记述了软件的早期发展，但在此之前这个主题偶尔也会出现，在后面的章节中，它出现得更频繁。在 20 世纪 50 年代，计算机公司将系统软件作为计算机价格的一部分提供，客户则开发自己的应用程序。不止一位早期计算系统的购买者对公司必须雇佣大批系统分析师、程序员和软件专家来管理一台本应消除文职人员的机器感到不满。直到 1990 年，随着硬件价格下降以及计算机系统变得更加可靠、紧凑和标准化，商业软件才在计算领域占据主导地位。

Another theme concerns a term, unknown in the late 1940s, that dominates computing in the 1990s: software.²² Chapter 3 chronicles the early development of software, but the topic crops up occasionally before that, and in the chapters that follow it appears with more frequency. In the 1950s computer companies supplied system software as part of the price of a computer, and customers developed their own applications programs. More than one purchaser of an early computing system winced at the army of systems analysts, programmers, and software specialists that had to be hired into the company to manage a machine that was supposed to eliminate clerical workers. It was not until 1990 that commercial software came to the fore of computing, as hardware prices dropped and computer systems became more reliable, compact, and standardized.

关于计算机历史的文献承认软件的重要性，但奇怪的是，这些文献被分成了两个阵营，似乎都没有认识到软件对另一个阵营的依赖。在一个阵营中，我们发现了大量关于个人计算机软件公司（尤其是微软）的书籍和杂志文章，以及销售个人计算机的 DOS 和 Windows 操作系统所赚到的钱。有些记载了 UNIX 的历史，UNIX 是一个有影响力的操作系统，在商业市场上也取得了成功。这些记载缺乏平衡。读者自然会对巨额的金钱交易感兴趣，但这个软件是做什么的，为什么操作系统看起来是这个样子？此外，这些记载中很少有将这些系统与计算机前二十年开发的软件联系起来，好像它们之间毫无关系。事实上，它们之间存在着密切的联系。

The literature on the history of computing recognizes the importance of software, but this literature is curiously divided into two camps, neither of which seems to recognize its dependence on the other. In one camp we find a glut of books and magazine articles about personal computer software companies, especially Microsoft, and the fortunes made in selling the DOS and Windows operating systems for PCs. Some chronicle the history of UNIX, an influential operating system that has also had success in the commercial marketplace. These accounts lack balance. Readers are naturally interested in the enormous sums of money changing hands, but what does this software do, and why do the operating systems look the way they do? Moreover, few of these chronicles connect these systems to the software developed in the first two decades of computing, as if they had nothing to do with each other. In fact, there are strong connections.

另一阵营则坚持更高的学术标准和客观性，并适当强调个人电脑出现之前的计算。但这些文献主要集中在编程语言上。从某种意义上说，这种方法反映了早期计算本身的活动，当时不难找到研究新编程语言和改进编程语言的人，但很难找到关心将这些语言集成到系统中以高效完成工作并充分利用客户时间的人。23^我们现在对 FORTRAN、BASIC、COBOL 和许多其他不太知名的语言的早期开发了解很多，但对这些语言所属的系统知之甚少。

Another camp adheres to higher standards of scholarship and objectivity, and gives appropriate emphasis to computing before the advent of the PC. But this body of literature has concentrated its efforts on programming languages. In a sense, this approach mirrors activity in computing itself in the early days, when it was not hard to find people working on new and improved programming languages, but was hard to find people who worried about integrating these languages into systems that got work done efficiently and made good use of a customer’s time.²³ We now know a great deal about the early development of FORTRAN, BASIC, COBOL, and a host of other more obscure languages, yet we know little of the systems those languages were a part of.

最后一个主题是信息在民主社会中的地位。计算机具有许多与印刷机相关的价值观，而印刷机的自由受到美国宪法第一修正案的保障。但计算机也是控制的媒介。24^这两个属性是否互相矛盾？商用计算机的第一批客户是军事或政府机构，他们希望这些机器能够管理使它们运作陷入瘫痪的信息；与此同时，大众媒体则将“自动化”吹捧为美国工人开启休闲和富裕新时代的媒介。一方面，旋风计划导致了 SAGE 的诞生，这是一个集中式指挥控制系统，其结构与资助它的空军的指挥结构相似；另一方面，它导致了数字设备公司的诞生，这家公司成立的目标是让计算机更便宜，让更多人更容易使用。个人电脑作为解放技术的概念将在第 7 章中详细讨论；我们要问的问题是，随着 20 世纪 80 年代个人电脑进入企业办公室，这些理想是否被扭曲了。我们还将看到，随着互联网在大众市场的迅猛发展，同样的问题再次浮出水面。

A final theme is the place of information in a democratic society. Computers share many values associated with the printing press, the freedom of which is guaranteed by the First Amendment to the U.S. Constitution. But computers are also agents of control.²⁴ Are the two attributes at odds with each other? The first customers for commercial computers were military or government agencies, who hoped these machines could manage the information that was paralyzing their operations; at the same time, the popular press was touting “automation” as the agent of a new era of leisure and affluence for American workers. Project Whirlwind led, on the one hand, to SAGE, a centralized command-and-control system whose structure mirrored the command structure of the Air Force, which funded it; on the other, it led to the Digital Equipment Corporation, a company founded with the goal of making computers cheaper and more accessible to more people. The notion of personal computers as liberating technology will be discussed in detail in chapter 7; the question we shall ask is whether those ideals were perverted as PCs found their way into corporate offices in the 1980s. We shall also see that these same issues have surfaced once again as the Internet has exploded into a mass market.

计算机软件是这一叙述的重要组成部分，但软件开发的一个方面被排除在外——人工智能 (AI)。人工智能探讨的是计算机是否可以执行一些任务，如果由人类完成，就会被普遍认为是智能的体现。机器智能最初被认为是硬件问题，但在这一历史的大部分时间里，人工智能研究都是通过编写程序来处理这个问题，这些程序运行在为其他应用程序制造和销售的相同存储程序数字计算机上（可能有一些增强功能）。人工智能涵盖的范围很广——从日常商业用途中相当平淡无奇的应用到关于人性的哲学问题。人工智能研究的定义在不断变化：例如，廉价的袖珍国际象棋机不是人工智能，但计算机进行的高级国际象棋仍然是人工智能。套用艾伦·图灵的话，人工智能的历史也许最好是由计算机而不是人来书写。

Computer software is very much a part of this narrative, but one facet of software development is excluded—Artificial Intelligence (AI). AI explores the question of whether computers can perform tasks that, if done by human beings, would be universally regarded as indicating intelligence. Machine intelligence was first believed to be a problem of hardware, but for most of this history AI research has dealt with it by writing programs that run on the same stored-program digital computers (perhaps with some enhancements) that are made and sold for other applications. Artificial Intelligence spans a wide range—from fairly prosaic applications in daily commercial use to philosophical questions about the nature of humanity. What defines AI research is constantly changing: cheap pocket chess-playing machines are not AI, for example, but advanced chess playing by computer still is. To paraphrase Alan Turing, the history of AI is perhaps better written by a computer than by a person.

本书着重介绍计算机在美国的发展历史。西欧，尤其是英国，也是电子计算机的先驱制造地，最初用于军事，后来用于商业。到 1943 年 12 月，当 ENIAC 才刚刚开始建造时，英国人已经拥有至少一台电子“巨人”投入使用。英国餐饮公司 J. Lyons & Company 在美国 UNIVAC 找到第一批客户之前就安装并使用了商用计算机 LEO。25^在德国，康拉德·楚泽 (Konrad Zuse) 也在采取措施将他在二战期间为德国军方创造的发明商业化。然而，到 20 世纪 50 年代末，欧洲人所拥有的领先优势都被美国公司夺走了。经济学家肯尼斯·弗拉姆 (Kenneth Flamm) 提出了一个原因：“20 世纪 50 年代，欧洲政府只提供了有限的资金来支持电子元件和计算机技术的开发，并且不愿意购买新的和未经试验的技术用于军事和其他系统。” ²⁶在欧洲，军用计算机和商用计算机之间几乎没有信息流通，更重要的是，人员流通也非常不顺畅。下文将偶尔介绍欧洲的贡献，但出于篇幅原因，不会详述欧洲计算机行业的发展历程。

This book focuses on the history of computing as it unfolded in the United States. Western Europe, especially England, was also a site where pioneering electronic computing machines were built, first for military and then for commercial customers. By December 1943, when construction of the ENIAC had only begun, the British already had at least one electronic “Colossus” in operation. The British catering firm J. Lyons & Company had installed and was using a commercial computer, the LEO, well before the American UNIVACs found their first customers.²⁵ In Germany, Konrad Zuse was also taking steps to commercialize inventions he had created for the German military during World War II. By the late 1950s, though, whatever lead the Europeans had was lost to American companies. The economist Kenneth Flamm suggests one reason for this: “European governments provided only limited funds to support the development of both electronic component and computer technology in the 1950s and were reluctant to purchase new and untried technology for use in their military and other systems.”²⁶ There was little of the easy flow of information—and more important, people—between military and commercial computing in Europe. The following narrative will occassionally address European contributions, but for reasons of space will not chronicle the unfolding of the computer industry there.

本文也将略微涉及日本的计算机历史。日本的历史则有所不同：日本在计算机领域起步较晚，从未生产过真空管计算机。然而，日本公司在集成电路生产方面取得了令人瞩目的进步，到 20 世纪 80 年代已在部分行业中站稳了脚跟。20 世纪 80 年代初，日本宣布了“第五代”计划，旨在超越美国的软件专业技术，此举在美国引起了很大不安，但美国在 20 世纪 90 年代仍保持了软件领域的领先地位。27^第5 章简要讨论了日本公司如何站稳脚跟。

This narrative will also touch only lightly on the history of computing in Japan. That story is different: Japan had a late start in computing, never producing vacuum tube computers at all. Japanese firms made remarkable advances in integrated circuit production, however, and had established a solid place in portions of the industry by the 1980s. The announcement in the early 1980s of a Japanese “Fifth Generation” program, intended to leapfrog over U.S. software expertise, created a lot of anxiety in the United States, but the United States retained its leadership in software into the 1990s.²⁷ How Japanese firms gained a foothold is discussed briefly in chapter 5.

冷战的结束以及苏联档案的开放可能有助于我们更好地了解苏联计算机的发展。在整个时期，苏联在计算机领域远远落后于美国。28 到目前为止，^给出的理由往往是事后诸葛亮：因为是这样，所以不得不这样。但苏联在纯数学和物理学以及开发弹道导弹、核武器、太空探索和超音速飞机方面的成就又如何呢？人们可能认为，苏联军方支持计算机的原因与美国空军支持 Whirlwind 和 SAGE 的原因相同。我们知道，苏联科学家在 ENIAC 发布后立即开始研究先进的数字计算机。然而，当他们需要先进的机器时，苏联人就会转向他们的东欧卫星国（特别是匈牙利和捷克斯洛伐克），或者对美国计算机（如 IBM System/360 和 VAX）进行逆向工程。通过复制这些计算机，他们可以访问大量软件，他们可以通过在公开市场上购买或通过间谍活动获得这些软件，但这也意味着他们的硬件技术仍然比美国落后一代或两代。

The end of the Cold War, and with it the opening of Soviet archives, may help us better understand the development of computing in the U.S.S.R. Throughout this era the Soviets remained well behind the United States in computing.²⁸ So far the reasons that have been given tend to be post hoc: because it was so, therefore it had to be so. But what of Soviet achievements in pure mathematics and physics, as well as in developing ballistic missiles, nuclear weapons, space exploration, and supersonic aircraft? One might expect that the Soviet military would have supported computing for the same reasons the U.S. Air Force supported Whirlwind and SAGE. We know that Soviet scientists began work on advanced digital computers as soon as the ENIAC was publicized. Yet when they needed advanced machines, the Soviets turned to their East European satellites (especially Hungary and Czechoslovakia), or else they reverse-engineered U.S. computers such as the IBM System/ 360 and the VAX. Building copies of these computers gave them access to vast quantities of software, which they could acquire by a purchase on the open market, or by espionage, but it also meant that they remained one or two hardware generations behind the United States.

或许是因为人们认为计算机是促进信息自由交换的工具，与极权主义国家是对立的。但从 1945 年到 1970 年代，美国的计算机系统由大型、集中式系统主导，并受到严格控制，而这些系统与苏联的政治制度并不矛盾。这种计算机将是模拟马克思列宁主义计划经济的完美工具。苏联的规划者并不孤单。在整个这个时代，一些美国人接受了计算机，因为它们有潜力为美国进行集中式经济建模——当然，宪法权利也得到了保障。29^原因可能与西欧相反：大量的军事支持，但没有转移到市场驱动的计算机行业。美国人可能发现军事支持“恰到好处”：足以支持创新，但又不会过于专注于特定的武器系统而扼杀创造力。需要对苏联计算机的历史进行更多的研究。

Perhaps it was the perception that computers, being instruments that facilitate the free exchange of information, are antithetical to a totalitarian state. But U.S. computing from 1945 through the 1970s was dominated by large, centralized systems under tight controls, and these were not at odds with the Soviet political system. Such computers would have been perfect tools to model the command economy of Marxism-Leninism. Soviet planners would not have been alone. Throughout this era some Americans embraced computers for their potential to perform centralized economic modeling for the United States—with constitutional rights guaranteed, of course.²⁹ Perhaps the reason was the other side of the Western European coin: plenty of military support, but no transfer to a market-driven computer industry. Americans may have found that military support was “just right”: enough to support innovation but not so focused on specific weapons systems as to choke off creativity. More research on the history of Soviet computing needs to be done.

我们大多数人都知道，计算机在影响现代生活的方式上与洗衣机有所不同。本书最后总结了一些关于为什么会这样的看法。在整个叙述过程中，我质疑计算机本身是否是变革的非人为推动者，甚至质疑它是否是一种自主的力量，人们几乎无法影响它，更不用说抵抗它了。在我的结论中，我重新审视了这个问题。我没有答案。我希望这些章节中呈现的编年史能够启发那些继续提问的人，无论是外行还是专业人士。

Most of us know that computers are somehow different from washing machines in the ways they are affecting modern life. This book concludes with some observations about why that might be so. Throughout the narrative I question whether the computer is itself the impersonal agent of change, or even whether it is an autonomous force that people can do little to affect, much less resist. In my conclusion I revisit that question. I do not have an answer. My hope is that the chronicle presented in these chapters will enlighten those of us, lay and professional, who continue to ask.

商业计算的出现，1945-1956 年

The Advent of Commercial Computing, 1945–1956

“你们这些家伙应该回去彻底修改你们的程序，别再跟 Eckert 和 Mauchly 搞这种愚蠢的勾当了。”这是哈佛数学家、Mark I 计算器的发明者 Howard Aiken 1948 年向美国国家标准局的 Edward Cannon 表达的观点。Aiken 是作为国家研究委员会成员发表上述言论的，该委员会刚刚建议标准局不要支持 J. Presper Eckert 和 John Mauchly 制造和销售电子计算机的提议（图 1.1 ^）。在 Aiken 看来，这种计算机永远不会形成商业市场；在美国，可能需要五六台这样的计算机，但不会更多。1

“[Y]ou … fellows ought to go back and change your program entirely, stop this … foolishness with Eckert and Mauchly.” That was the opinion of Howard Aiken, Harvard mathematician and builder of the Mark I calculator, expressed to Edward Cannon of the U.S. National Bureau of Standards in 1948. Aiken made that remark as a member of a National Research Council committee that had just recommended that the Bureau of Standards not support J. Presper Eckert and John Mauchly’s proposal to make and sell electronic computers (figure 1.1). In Aiken’s view, a commercial market would never develop; in the United States there was a need for perhaps for five or six such machines, but no more.¹

霍华德·艾肯错了。到 20 世纪 90 年代，市场对数百万台电子数字计算机的需求已经显现，其中许多是可以轻松放入公文包的个人设备。如果没有固体物理学的进步，这种情况就不会发生，固体物理学提供了一种将计算机电路放在几块硅片上的方法。尽管如此，20 世纪 90 年代几乎无处不在的计算机是埃克特和莫奇利在 20 世纪 40 年代末希望商业化的计算机的直接后代。

Howard Aiken was wrong. There turned out to be a market for millions of electronic digital computers by the 1990s, many of them personal devices that fit easily into a briefcase. That would not have happened were it not for advances in solid state physics, which provided a way of putting the circuits of a computer on a few chips of silicon. Nevertheless, the nearly ubiquitous computers of the 1990s are direct descendants of what Eckert and Mauchly hoped to commercialize in the late 1940s.

埃克特-莫奇利计算机公司并没有长期保持独立实体地位，它被雷明顿兰德公司收购，成为雷明顿兰德公司商业机器公司的一个部门。埃克特和莫奇利的计算机 UNIVAC 是一款技术杰作，但在市场上却被雷明顿兰德公司的竞争对手 IBM 制造的计算机所取代。因此，可以说，他们确实愚蠢地低估了将他们的发明商业化的难度。他们并不愚蠢的是他们的远见，他们不仅知道如何设计和制造计算机，还知道如何让社会从大量计算机中受益。

The Eckert-Mauchly Computer Corporation did not remain an independent entity for long; it was absorbed by Remington Rand and became a division of that business-machine company. Eckert and Mauchly’s computer, the UNIVAC, was a technical masterpiece but was eclipsed in the market by computers made by Remington-Rand’s competitor, IBM. So one could say that they were indeed foolish in their underestimation of the difficulties of commercializing their invention. What was not foolish was their vision, not only of how to design and build a computer but also of how a society might benefit from large numbers of them.

图 1.1

1948 年左右，费城 Eckert-Mauchly 计算机公司的员工。左下方是 Eckert，右下方是 Mauchly。他们身后的仪器是 BINAC 的一部分，该公司正在为 Northrop Aircraft Company 制造 BINAC。后排，从左到右： Albert Auerbach、Jean Bartik、Marvin Jacoby、John Sims、Louis Wilson、Robert Shaw、Gerald Smoliar。前排：J. Presper Eckert、Frazier Welsh、James Wiener、Bradford Sheppard、John Mauchly。（来源：Unisys Corporation。）

Figure 1.1

Staff of the Eckert-Mauchly Computer Corporation, ca. 1948, in Philadelphia. Eckert is at the lower left; Mauchly at the lower right. The apparatus behind them is a portion of the BINAC, which the company was building for the Northrop Aircraft Company. Back row, left to right : Albert Auerbach, Jean Bartik, Marvin Jacoby, John Sims, Louis Wilson, Robert Shaw, Gerald Smoliar. Front row : J. Presper Eckert, Frazier Welsh, James Wiener, Bradford Sheppard, John Mauchly. (Source : Unisys Corporation.)

1945 年以后的计算是一群人在关键时刻重新定义技术本身性质的故事。他们这样做，将计算推向了新的市场、新的应用，并在社会秩序中占据了新的地位。埃克特和莫奇利是众多实现这种转变的人中的第一批。他们将一种昂贵而脆弱的科学仪器（类似于回旋加速器）变成了一种可以制造和销售的产品，尽管数量很少。2 ²⁰世纪 50 年代中期，IBM 公司开发了一系列产品，满足了美国企业的信息处理需求。十年后，麻省理工学院旋风项目的校友将计算机变成了一种可以与之交互的设备，一种增强智力的工具。20 世纪 70 年代中期，一群业余爱好者和发烧友将其变成了个人设备。1980 年左右，它从一种专用硬件转变为一种由其现已商业化的软件定义的标准化消费产品。 20 世纪 90 年代，计算机又经历了一次转型，成为全球联系的载体和通信媒介。“计算机时代”——实际上是一系列“计算机时代”——并非偶然被发明出来；它是人们希望它发生时才出现的。这一再发明和重新定义的过程仍在继续。

Computing after 1945 is a story of people who at critical moments redefined the nature of the technology itself. In doing so they opened up computing to new markets, new applications, and a new place in the social order. Eckert and Mauchly were the first of many who effected such a transformation. They took an expensive and fragile scientific instrument, similar to a cyclotron, and turned it into a product that could be manufactured and sold, if only in small quantities.² In the mid-1950s the IBM Corporation developed a line of products that met the information-handling needs of American businesses. A decade later, alumni from MIT’s Project Whirlwind turned the computer into a device that one interacted with, a tool with which to augment one’s intellectual efforts. In the mid-1970s, a group of hobbyists and enthusiasts transformed it into a personal appliance. Around 1980, it was transformed from a piece of specialized hardware to a standardized consumer product defined by its now-commercialized software. In the 1990s it is going through another transformation, turning into an agent of a worldwide nexus, a communications medium. The “computer age”— really a series of “computer ages”—was not just invented; it was willed into existence by people who wanted it to happen. This process of reinvention and redefinition is still going on.

UNIVAC 背景

The UNIVAC in Context

1951 年，埃克特和莫奇利发明了一台名为“UNIVAC”的计算机，这是他们第一次进行这样的变革。这个缩写词来自“通用自动计算机”，这是他们精心挑选的名字。“通用”意味着它可以解决科学家、工程师和企业遇到的问题。“自动”意味着它可以解决复杂问题，而不需要像现有技术那样不断进行人为干预或判断。在讨论它的发明之前，人们需要了解不同领域的计算工作是如何进行的，以及为什么一台机器——UNIVAC——可以同样出色地为它们服务。人们还必须了解现有的计算机——几十年来改进和使用的成果——存在哪些不足之处。正是这种不足为 UNIVAC 创造了空间，它在许多方面打破了过去的做法。

Eckert and Mauchly brought on the first of these transformations in 1951 with a computer they called “UNIVAC.” The acronym came from “Universal Automatic Computer,” a name that they chose carefully. “Universal” implied that it could solve problems encountered by scientists, engineers, and businesses. “Automatic” implied that it could solve complex problems without requiring constant human intervention or judgment, as existing techniques required. Before discussing its creation, one needs to understand how computing work was being done in different areas and why a single machine, a UNIVAC, could serve them equally well. One must also understand how existing calculating machines, the results of decades of refinement and use, were deficient. It was that deficiency that made room for the UNIVAC, which broke with past practices in many ways.

打孔卡

Punched Cards

第二次世界大战期间，埃克特和莫奇利在宾夕法尼亚大学摩尔电气工程学院设计并制造了 ENIAC。ENIAC 是一款电子计算器，开创了美国数字计算的时代。它的目的是为美国陆军计算射击表，这项任务涉及重复求解复杂的数学表达式。正是在研究这种设备时，他们构想出了一种更具普遍吸引力的东西。

During the Second World War, Eckert and Mauchly designed and built the ENIAC at the University of Pennsylvania’s Moore School of Electrical Engineering. The ENIAC was an electronic calculator that inaugurated the era of digital computing in the United States. Its purpose was to calculate firing tables for the U.S. Army, a task that involved the repetitive solution of complex mathematical expressions. It was while working on this device that they conceived of something that had a more universal appeal.

UNIVAC 中的信息流反映了 Eckert 和 Mauchly 的物理学和工程学背景。也就是说，UNIVAC 中的指令和数据流反映了人类使用机械计算器、表格簿以及铅笔和纸张进行科学计算的方式。3^尽管真空管电路可能看起来很新颖，但科学家或工程师不会发现 UNIVAC 解决问题的方式有什么不寻常之处。

The flow of information through the UNIVAC reflected Eckert and Mauchly’s background in physics and engineering. That is, the flow of instructions and data in the UNIVAC mirrored the way humans using mechanical calculators, books of tables, and pencil and paper performed scientific calculations.³ Although the vacuum tube circuits might have appeared novel, a scientist or engineer would not have found anything unusual in the way a UNIVAC attacked a problem.

然而，那些从事商业计算的人（Eckert 和 Mauchly 的客户也希望他们的机器能为他们服务）会发现 UNIVAC 的处理方法很不寻常。4^十九^世纪末，许多企业采用了使用打孔卡机来组织工作的做法；通常，一套装置由三到六种不同的打孔卡设备组成。5要用计算机取代这些机器，企业还必须采用 UNIVAC 的信息处理方式。打孔卡机通常被称为“单位记录设备”。有了它们，关于特定实体（例如，销售交易）的所有相关信息都编码在一张卡片上，这张卡片可以通过不同的设备运行而实现多种用途；例如，计数、排序、制表或打印特定的一组列。6^打孔卡机的历史记载非常详细描述了各个机器的功能。更相关的是整个房间的“建筑”——包括里面的人——包括一个打孔卡装置，因为电子计算机最终取代的是那个房间，而不是单个的机器。

However, those engaged in business calculations, customers Eckert and Mauchly also wanted their machine to serve, would have found the UNIVAC’s method of processing unusual.⁴ In the late nineteenth century, many businesses adopted a practice that organized work using a punched card machine; typically an ensemble of three to six different punched-card devices would comprise an installation.⁵ To replace these machines with a computer, the business had also to adopt the UNIVAC’s way of processing information. Punched-card machines are often called “unit record equipment.” With them, all relevant information about a particular entity (e.g., a sales transaction) is encoded on a single card that can serve multiple uses by being run through different pieces of equipment; for example, to count, sort, tabulate, or print on a particular set of columns.⁶ Historical accounts of punched-card machinery have described in great detail the functioning of the individual machines. More relevant is the “architecture” of the entire room—including the people in it—that comprised a punched-card installation, since it was that room, not the individual machines, that the electronic computer eventually replaced.

在典型的穿孔卡安装中，当一副穿孔卡经过制表机或其他机器时，对文件中的所有记录执行相同的操作（图 1.2）。UNIVAC 及其后继者可以这样操作，但它们也可以在从内存中获取下一个记录之前对单个数据执行一长串操作。用穿孔卡的术语来说，这就要求带着一“副”穿孔卡在房间里走来走去——这对机器或人力来说都不是很经济。到 20 世纪 30 年代中期，对收集到的一副穿孔卡中的信息进行处理已成为商业实践的一部分，而穿孔卡设备销售人员深入客户会计办公室的做法也进一步巩固了这一做法^。7

In a typical punched-card installation, the same operation was performed on all the records in a file as a deck of cards went through a tabulator or other machine (figure 1.2). The UNIVAC and its successors could operate that way, but they could also perform a long sequence of operations on a single datum before fetching the next record from memory. In punched-card terms, that would require carrying a “deck” of a single card around the room—hardly an economical use of the machinery or the people. Processing information gathered into a deck of cards was entrenched into business practices by the mid-1930s, and reinforced by the deep penetration of the punched-card equipment salesmen into the accounting offices of their customers.⁷

到了 20 世纪 30 年代，一些科学家，特别是天文学家，开始使用打孔卡设备解决科学问题。他们发现对每个数据执行一系列操作很有意义，因为下一个操作通常取决于前一个操作的结果。其中一位是 Wallace Eckert（与 J. Presper Eckert 无关），他在 IBM 的帮助下于 1934 年在纽约哥伦比亚大学建立了 Thomas J. Watson 计算局。1940 年，他在一本颇具影响力的著作《科学计算中的打孔卡方法》中总结了他的工作。在书中，他解释说，打孔卡机“都是为计算而设计的，每个操作在开始下一个操作之前都要在许多卡片上完成。” ⁸他强调了如何使用现有设备进行科学工作，但他表示，建造专门的机器来解决科学问题不值得“花费大量金钱和时间”。⁹十年后，J. Presper Eckert 和 John Mauchly 提议做的事正是如此——不惜花费巨资和精力，创造一种能够同时处理商业和科学问题的“通用”机器。

By the 1930s a few scientists, in particular astronomers, began using punched-card equipment for scientific problems. They found that it made sense to perform sequences of operations on each datum, since often the next operation depended on the results of the previous one. One such person was Wallace Eckert (no relation to J. Presper Eckert), who with the aid of IBM established the Thomas J. Watson Computing Bureau at Columbia University in New York in 1934. In 1940 he summarized his work in an influential book, Punched Card Methods in Scientific Computation. In it, he explained that punched-card machines “are all designed for computation where each operation is done on many cards before the next operation is begun.”⁸ He emphasized how one could use existing equipment to do scientific work, but he stated that it was not worth the “expense and delay involved” in building specialized machines to solve scientific problems.⁹ A decade later, that was precisely what J. Presper Eckert and John Mauchly were proposing to do—go to great expense and effort to create a “universal” machine that could handle both business and scientific problems.

图 1.2

IBM 打孔卡。摘自 IBM 公司《IBM 数据处理功能》手册 224-8208-5，约 1963 年。（来源：IBM 公司。）

Figure 1.2

IBM punched card. From IBM Corporation, “IBM Data Processing Functions,” Brochure 224-8208-5, ca. 1963. (Source : IBM Corporation.)

讽刺的是，华莱士·埃克特 (Wallace Eckert) 是第一批放弃传统打孔卡实践、转向后来出现的数字计算机的人之一。尽管他建议不要制造专门的设备，但他的实验室确实设计了一种称为控制开关的设备。他将这个开关安装在乘法器、制表机和汇总打孔机之间。它的功能是允许在读取下一张卡片之前在一张卡片上执行短序列操作（最多 12 个）。10^根据他的建议，IBM 在马里兰州阿伯丁的美国陆军弹道研究实验室建造并安装了两台专门制造的打孔卡机。IBM 将这些机器称为“阿伯丁继电器计算器”；它们后来被称为 PSRC，即“可插拔序列继电器计算器” ^。11

Ironically, Wallace Eckert was among the first to venture away from traditional punched-card practices and toward one more like the digital computers that would later appear. Despite his recommendation against building specialized equipment, he did have a device called a control switch designed at his laboratory. He installed this switch between the multiplier, tabulator, and summary punch. Its function was to allow short sequences of operations (up to 12) to be performed on a single card before the next card was read.¹⁰ Following his advice, IBM built and installed two specially built punched-card machines at the U.S. Army’s Ballistic Research Laboratory at Aberdeen, Maryland. IBM called these machines the “Aberdeen Relay Calculators”; they were later known as the PSRC, for “Pluggable Sequence Relay Calculator.”¹¹

1945 年末，又为其他军事实验室建造了三台继电器计算器，这些计算器更加复杂。在一台这样的计算器读取卡片的时间内，它可以执行多达 48 个步骤的序列。还可以执行更复杂的序列内序列。12^一位计算机科学家后来指出，这种编程方法需要“如今在某些计算机的微编程级别上看到的那种并行子序列的详细设计”。13^如果编程正确，这些计算器比任何其他非电子计算器都快。即使在 ENIAC 完成安装并从费城搬到阿伯丁之后，弹道研究实验室还建造了更多的继电器计算器。它们在 1952 年仍在使用，那时 BRL 不仅拥有 ENIAC，还拥有 EDVAC、ORDVAC（均为电子计算机）、IBM 卡片编程计算器（下文将介绍）和贝尔实验室 Model V（一种非常大的可编程继电器计算器）^。14

In late 1945, three more were built for other military labs, and these were even more complex. During the time one of these machines read a card, it could execute a sequence of up to forty-eight steps. More complex sequences-within-sequences were also possible.¹² One computer scientist later noted that this method of programming demanded “the kind of detailed design of parallel subsequencing that one sees nowadays at the microprogramming level of some computers.”¹³ When properly programmed, the machines were faster than any other nonelectronic calculator. Even after the ENIAC was completed and installed and moved from Philadelphia to Aberdeen, the Ballistic Research Lab had additional Relay Calculators built. They were still in use in 1952, by which time the BRL not only had the ENIAC but also the EDVAC, the ORDVAC (both electronic computers), an IBM Card Programmed Calculator (described next), and the Bell Labs Model V, a very large programmable relay calculator.¹⁴

卡片编程计算器

The Card-Programmed Calculator

阿伯丁继电器计算器从未成为商业产品，但它们揭示了尝试将现有设备改造为二战后的需求，而不是采取革命性的方法，例如 UNIVAC。还有其他穿孔卡设备代表了 Eckert 和 Mauchly 提出的发明的真正商业替代品。1935 年，IBM 推出了乘法穿孔机（型号 601）；这些穿孔卡很快就成为科学或统计工作的流行产品。1946 年，IBM 推出了改进型号 603，这是第一款使用真空管进行计算的 IBM 商业产品。两年后，IBM 用 604 取代了它，它不仅使用了真空管，还采用了阿伯丁机器开创的排序功能。除了其他穿孔卡设备常见的插线板控制外，它还可以为每次读取卡片和设置插线板执行最多 60 个步骤。¹⁵ 604 及其后继产品 IBM 605 成为许多设施科学计算的支柱，直到 20 世纪 50 年代中期可靠的商用计算机问世。它是 IBM 在那个时代最成功的产品之一：1948 年至 1958 年间生产了 5,000 多台^。16

The Aberdeen Relay Calculators never became a commercial product, but they reveal an attempt to adapt existing equipment to post–World War II needs, rather than take a revolutionary approach, such as the UNIVAC. There were also other punched-card devices that represented genuine commercial alternatives to Eckert and Mauchly’s proposed invention. In 1935 IBM introduced a multiplying punch (the Model 601); these soon became popular for scientific or statistical work. In 1946 IBM introduced an improved model, the 603, the first commercial IBM product to use vacuum tubes for calculating. Two years later IBM replaced it with the 604, which not only used tubes but also incorporated the sequencing capability pioneered by the Aberdeen machines. Besides the usual plugboard control common to other punched-card equipment, it could execute up to 60 steps for each reading of a card and setting of the plugboard.¹⁵ The 604 and its successor, the IBM 605, became the mainstays of scientific computing at many installations until reliable commercial computers became available in the mid 1950s. It was one of IBM’s most successful products during that era: over 5,000 were built between 1948 and 1958.¹⁶

IBM 最大的工程客户之一、位于加利福尼亚州霍桑的 Northrop Aircraft 公司将一台 603 乘法打孔机连接到他们的一台制表机上。这样 Northrop 的用户就可以将计算结果打印在纸上，而不是打孔到卡片上。通过进一步的轻微修改和添加一个小盒子，将数字存储在继电器组中，该机器可以使用通过制表机的打孔卡来控制乘法器执行的序列^。17

One of IBM’s biggest engineering customers, Northrop Aircraft of Hawthorne, California, connected a 603 multiplying punch to one of their tabulating machines. That allowed Northrop’s users to print the results of a calculation on paper instead of punching them on cards. With a slight further modification and the addition of a small box that stored numbers in banks of relays, the machine could use punched cards run through the tabulator to control the sequences carried out by the multiplier.¹⁷

从逻辑上讲，这种布置与普通的打孔卡装置没什么不同，只是一组电缆和控制箱取代了将卡片从一台机器运送到另一台机器的人员。诺斯罗普的一位工程师多年后回忆说，他们之所以特意安排这种布置，是因为他们正在解决一个问题，该问题的下一步取决于上一步的结果。这意味着，通过机器运行的正常卡片组减少为“一批 [卡片]，这很不方便”。18^换句话说，有了电缆连接机器，装置就变成了按顺序执行指令的装置，而且比插入电缆更灵活地进行编程。

Logically, the arrangement was no different from an ordinary punched card installation, except that a set of cables and control boxes replaced the person whose job had been to carry decks of cards from one machine to the next. One of the Northrop engineers recalled years later that they rigged up the arrangement because they were running a problem whose next step depended on the results of the previous step. What this meant was that the normal decks of cards that ran through a machine were reduced to “a batch of one [card], which was awkward.”¹⁸ In other words, with cables connecting the machines, the installation became one that executed instructions sequentially and was programmable in a more flexible way than plugging cables.

IBM 后来将该组合的一个版本作为卡片编程计算器 (CPC) 进行销售。19^从1948 年到 1950 年代中期，总共安装了数百台计算机——远远少于以传统方式安装的数千台制表机、打孔机和其他设备。但即使是这个数字也比 1954 年左右全球安装的电子计算机数量多出许多倍。对于像 Northrop 这样的工程导向型公司来说，CPC 满足了迫切的需求，而这些需求不能等到与营销存储程序计算机相关的问题得到解决^。20

IBM later marketed a version of this ensemble as the Card-Programmed Calculator (CPC).¹⁹ Perhaps several hundred in all were installed between 1948 and the mid 1950s—far fewer than the thousands of tabulators, punches, and other equipment installed in the traditional way. But even that was many times greater than the number of electronic computer installations worldwide until about 1954. For engineering oriented companies like Northrop, the CPC filled a pressing need that could not wait for the problems associated with marketing stored-program computers to be resolved.²⁰

阿伯丁计算器和 604 是计算器、制表机和 UNIVAC 等真正的计算机之间的过渡机器。CPC 过于依赖打孔卡方法，对计算机设计师来说毫无价值。到推出时，人们已经清楚地认识到 UNIVAC 所采用的设计更胜一筹，在这种设计中，指令和数据都存储在内部存储设备中。卡片编程计算器的程序卡、插板和互连电缆的组合就像托勒密宇宙学后期迭代的本轮，而哥白尼体系已经开始被接受。21^然而，需要解决困难工程问题的客户接受了它。它比当时提供的计算机便宜，而且可以买到。除了诺斯罗普之外，其他南加州航空航天公司也仔细评估了卡片编程计算器与供应商对电子计算机的宣传。22^几乎所有公司都安装了至少一台 CPC。

The Aberdeen calculators and the 604 were transitional machines, between calculators, tabulators, and genuine computers like the UNIVAC. The CPC carried the punched-card approach too far to be of value to computer designers. By the time of its introduction, it was already clear that the design used by the UNIVAC, in which both the instructions and the data were stored in an internal memory device, was superior. The Card-Programmed Calculator’s combination of program cards, plugboards, and interconnecting cables was like the epicycles of a late iteration of Ptolemaic cosmology, while the Copernican system was already gaining acceptance.²¹ Customers needing to solve difficult engineering problems, however, accepted it. It cost less than the computers then being offered, and it was available. Other southern California aerospace firms besides Northrop carefully evaluated the Card-Programmed Calculator against vendors’ claims for electronic computers.²² Nearly all of them installed at least one CPC.

存储程序原理

The Stored-Program Principle

任何见过 UNIVAC 的人都会发现它与现有的计算器和打孔卡设备有很大不同。它使用真空管——有数千个。它将数据存储在磁带上，而不是卡片上。它是一个庞大而昂贵的系统，而不是不同设备的集合。最大的不同是它的内部设计，普通观察者是看不到的。UNIVAC 是一台“存储程序”计算机，是最早的计算机之一。最重要的是，这使它不同于它要取代的机器。

No one who saw a UNIVAC failed to see how much it differed from existing calculators and punched card equipment. It used vacuum tubes—thousands of them. It stored data on tape, not cards. It was a large and expensive system, not a collection of different devices. The biggest difference was its internal design, not visible to the casual observer. The UNIVAC was a “stored program” computer, one of the first. More than anything else, that made it different from the machines it was designed to replace.

在战时秘密中隐藏着将计算机程序存储在内部的想法。这个想法是在埃克特、莫奇利和其他人急于完成 ENIAC 以协助美国陆军时产生的，当时美国陆军正在欧洲和北非进行地面战争。这个想法的产生是因为 ENIAC 的创造者意识到，虽然 ENIAC 可能会正常工作，但它将是一台难以操作的机器。

The origins of the notion of storing a computer’s programs internally are clouded in war-time secrecy. The notion arose as Eckert, Mauchly, and others were rushing to finish the ENIAC to assist the U.S. Army, which was engaged in a ground war in Europe and North Africa. It arose because the ENIAC’s creators recognized that while the ENIAC was probably going to work, it was going to be a difficult machine to operate.

将现代术语“编程”应用于计算机可能起源于摩尔学院的 ENIAC 团队。不过，他们更常使用短语“设置”来描述配置 ENIAC 以解决不同问题。23设置 ENIAC 意味着插拔错综复杂的电缆并设置开关阵列。实际上，这台机器必须为解决每个新问题而重建。当 ENIAC 于 1945 年底完成时，^它的运行速度比之前的任何其他机器都要快得多。但是，虽然它可以在几秒钟内解决一个复杂的数学问题，但正确设置机器以做到这一点可能需要几天的时间。

Applying the modern term “to program” to a computer probably originated with the ENIAC team at the Moore School. More often, though, they used the phrase “set up” to describe configuring the ENIAC to solve different problems.²³ Setting up the ENIAC meant plugging and unplugging a maze of cables and setting arrays of switches. In effect, the machine had to be rebuilt for each new problem it was to solve. When completed in late 1945, the ENIAC operated much faster than any other machine before it. But while it could solve a complex mathematical problem in seconds, it might take days to set up the machine properly to do that.

正是在建造这台机器的过程中，它的创造者们构想出了一种替代方案。将这种见解融入 ENIAC 已经太晚了，但它确实为后续机器“EDVAC”（电子离散变量计算机）奠定了基础。在 1945 年 9 月撰写的描述中，埃克特和莫奇利简明扼要地阐述了这一概念：“这种设备的一个重要特点是操作指令和函数表将存储在与用于数字的存储设备完全相同的存储设备中。” ²⁴六个月后，埃克特和莫奇利离开了摩尔学院，EDVAC 的工作被移交给其他人（这主要是为什么花了五年时间才完成建造）。将指令和数据存储在一个公共存储单元中的概念将成为 UNIVAC 以及几乎所有后续计算机的基本特征。²⁵

It was in the midst of building this machine that its creators conceived of an alternative. It was too late to incorporate that insight into the ENIAC, but it did form the basis for a proposed follow-on machine called the “EDVAC” (Electronic Discrete Variable Computer). In a description written in September of 1945, Eckert and Mauchly stated the concept succinctly: “An important feature of this device was that operating instructions and function tables would be stored exactly in the same sort of memory device as that used for numbers.”²⁴ Six months later, Eckert and Mauchly left the Moore School, and work on the EDVAC was turned over to others (which was mainly why it took five more years to finish building it). The concept of storing both instructions and data in a common storage unit would become basic features of the UNIVAC and nearly every computer that followed.²⁵

存储程序原理是 UNIVAC 成功的关键。它首先让 Eckert 和 Mauchly 制造出一台比 ENIAC 具有更多通用功能的计算机，但需要的真空管却更少。它导致了“编程”（后来的“软件”）的确立，它既独立于硬件设计，又与硬件设计同等重要。在 1945 年至 1995 年的计算发展过程中，这种设计的基础保持了相当的稳定。直到这一时期的末期，我们才遇到与它有重大偏差的设计，即“大规模并行”处理器或“非冯·诺依曼”架构。

The stored-program principle was a key to the UNIVAC’s success. It allowed Eckert and Mauchly, first of all, to build a computer that had much more general capabilities than the ENIAC, yet required fewer vacuum tubes. It led to the establishment of “programming” (later “software”) as something both separate from and as important as hardware design. The basics of this design remained remarkably stable during the evolution of computing from 1945 to 1995. Only toward the end of this period do we encounter significant deviations from it, in the form of “massively parallel” processors or “non–von Neumann” architectures.

约翰·冯·诺依曼的角色

John von Neumann’s Role

尽管埃克特和莫奇利早在 1944 年就意识到计算机需要存储器程序，但约翰·冯·诺依曼于 1945 年 6 月 30 日撰写的“EDVAC 报告初稿”却经常被引用为现代计算的奠基文献。26^从这份报告以及几年后冯·诺依曼与人合作撰写的一系列报告中，诞生了“冯·诺依曼架构”一词来描述这种设计。27^根据负责 ENIAC 项目的军官赫尔曼·戈德斯坦的说法，约翰·冯·诺依曼 (1903-1957) 于 1944 年夏天在马里兰州阿伯丁火车站偶然遇到 ENIAC。28^尽管冯·诺依曼参与了许多其他项目，包括原子弹的设计，但他对摩尔学院正在发生的事情非常感兴趣，因此结识了埃克特和莫奇利并参与了这个项目。

Although Eckert and Mauchly had realized as early as 1944 that computers would need to store the program, the “First Draft of a Report on the EDVAC,” by John von Neumann, dated June 30, 1945, is often cited as the founding document of modern computing.²⁶ From it, and a series of reports co-authored by von Neumann a few years later, comes the term “von Neumann Architecture” to describe such a design.²⁷ According to Herman Goldstine, an army officer assigned to the ENIAC project, John von Neumann (1903–1957) learned of the ENIAC from a chance meeting with him in the summer of 1944 at the Aberdeen, Maryland, railroad station.²⁸ Despite his involvement in many other projects, including the design of the atomic bomb, von Neumann was sufficiently intrigued by what was going on at the Moore School to have himself introduced to Eckert and Mauchly and brought onto the project.

当时，埃克特和莫奇利正忙于思考如何加快计算机的设置速度。29^一种可能性是使用穿孔纸带来输入指令，就像 20 世纪 40 年代的几台中继机一样，但这对于 ENIAC 的高速计算电路来说太慢了。卡片编程计算器使用的卡片组也是如此。用莫奇利的话来说，“只有高速输入指令，才能高速进行计算。” ³⁰

Eckert and Mauchly were at that time busy thinking of ways to improve the process of setting up a computer faster.²⁹ One possibility was to use perforated paper tape to feed instructions, as several relay machines of the 1940s did, but this was too slow for the high speeds of the ENIAC’s calculating circuits. So were the decks of cards used by the Card-Programmed Calculator. In Mauchly’s words, “calculations can be performed at high speed only if instructions are supplied at high speed.”³⁰

1944 年，在 ENIAC 的建造过程中，埃克特撰写了《磁力计算机的披露》，其中描述了使用“至少外缘由磁性合金制成的盘或鼓”来存储数字。31尽管该披露侧重于设计一种“速度更快、更简单，同时具有实用性、坚固性和易维修性”的机器的方法，但并未阐明后来被称为存储程序原理的设计概念。32^{冯·诺依曼}¹⁹⁴⁵年关于 EDVAC 的报告走得更远——它从逻辑结构而不是硬件结构的角度描述了机器。埃克特和莫奇利于 1945 年 9 月提交的备忘录简明扼要地阐述了这一原理：他们写道，指令和数字数据将存储在“完全相同的存储设备中” ^。33

In the midst of the ENIAC’s construction in 1944, Eckert wrote a “Disclosure of a Magnetic Calculating Machine,” in which he described the use of “[d]iscs or drums which have at least their outer edge made of a magnetic alloy” on which numbers can be stored.³¹ Although it focused on ways of designing a machine that was “speedier, simpler as well as providing features of utility, ruggedness and ease or repair,” the disclosure did not articulate the design concepts that later would become known as the stored-program principle.³² Von Neumann’s 1945 Report on the EDVAC went farther—it described a machine in terms of its logical structure rather than its hardware construction. The memorandum that Eckert and Mauchly submitted in September 1945, stated the principle succinctly: they wrote that instructions and numerical data would be stored “in exactly the same sort of memory device.”³³

从上述报告和备忘录的顺序来看，埃克特和莫奇利在 1944 年就构想出了一种类似存储程序原理的东西，但冯·诺依曼澄清了它，并以一种强有力的形式陈述了它。冯·诺依曼作为数学家的国际声誉也使这个想法比仅靠埃克特和莫奇利的想法更有影响力，他们两人在摩尔学院之外都不出名。虽然“冯·诺依曼架构”这个术语根深蒂固，难以取代，但埃克特和莫奇利从工程角度展示了对电子计算本质的深刻理解，他们应该得到同样的赞誉^。34

From the above sequence of reports and memorandums it appears that Eckert and Mauchly had conceived of something like a stored-program principle by 1944, but that it was von Neumann who clarified it and stated it in a form that gave it great force. Von Neumann’s international reputation as a mathematician also gave the idea more clout than it might have had coming solely from Eckert and Mauchly, neither of whom were well-known outside the Moore School. Although the term “von Neumann Architecture” is too entrenched to be supplanted, Eckert and Mauchly, who demonstrated such a deep understanding of the nature of electronic computing from an engineering perspective, deserve equal credit.³⁴

1946 年夏天，摩尔学院和美国军方联合举办了一门“电子数字计算机设计理论与技术”课程。这门课程表明，学院无法满足 ENIAC 公开发布后提出的大量信息请求。35^这一系列课程讲座和一两年后出现的油印报告牢固地确立了摩尔学院的计算机设计方法。很快就出现了基于这一概念的机器。到 1948 年中期，英国曼彻斯特大学的一台实验计算机正在运行测试程序。剑桥大学的莫里斯·威尔克斯 (Maurice Wilkes) 在他的 EDSAC 中实现了这一想法，并于 1949 年春天投入使用。埃克特和莫奇利在当年晚些时候完成了 BINAC。36^当然，UNIVAC 也将采用它。其他人将继续提出和制造替代设计的电子计算机，但在 1946 年夏天之后，计算的道路至少在理论上是清晰的。

In the summer of 1946, the Moore School and the U.S. military cosponsored a course on the “Theory and Techniques for Design of Electronic Digital Computers.” The course was a recognition of the school’s inability to accommodate the numerous requests for information following the public unveiling of the ENIAC.³⁵ That series of course lectures and the mimeographed reports that appeared a year or two later firmly established the Moore School’s approach to computer design. Machines soon appeared that were based on that concept. An experimental computer at the University of Manchester, England, was running test programs by mid-1948. Maurice Wilkes, of Cambridge University, implemented the idea in his EDSAC, operational in the spring of 1949. Eckert and Mauchly completed the BINAC later that year.³⁶ And of course the UNIVAC would also employ it. Others would continue to propose and build electronic computers of alternate designs, but after the summer of 1946, computing’s path, in theory at least, was clear.

冯·诺依曼架构及其意义

The von Neumann Architecture and Its Significance

在对 UNIVAC 进行描述之前，值得简单回顾一下冯·诺依曼在 1945 年的报告中所描述的架构的基本要素，特别是在过去半个世纪的计算机设计中一直保持稳定的那些方面。

Before providing a description of the UNIVAC, it is worth a brief look at the essentials of the architecture that von Neumann described in his 1945 report, especially those aspects of it that have remained stable through the past half-century of computer design.

除了程序的内部存储之外，冯·诺依曼计算机的一个主要特点是处理信息的单元与存储信息的单元是分开的。通常，这两个单元之间只有一个通道，所有信息传输都必须通过这个通道（所谓的冯·诺依曼瓶颈，稍后会详细介绍）。这一特点主要是出于工程原因：设计不需要对其内容进行算术运算的存储单元更容易。

Aside from the internal storage of programs, a major characteristic of a von Neumann computer is that the units that process information are separate from those that store it. Typically there is only a single channel between these two units, through which all transfers of information must go (the so-called von Neumann Bottleneck, about which more later). This feature arose primarily for engineering reasons: it was easier to design storage cells that did not also have to perform arithmetic on their contents.

主要特点是指令和数据存储在同一个存储设备中，任何数据都可以像其他数据一样快速地从中检索。这一概念源于这样的考虑：计算机的处理单元不应该闲置，等待下一个指令的传递。除此之外，指令与数据的比例通常因问题而异，因此为每个问题专门配备单独的昂贵存储设备是没有意义的。这种设计意味着人们可以将编码指令视为数据并对其执行操作，从而将其更改为另一条指令，但最初并没有完全理解这一点。为了说明这是如何首次实现的，UNIVAC 主存储器可以容纳多达 1,000 个“字”，这些字可以是数字（11 位数字加符号）、字符（每个字 12 个字符）或指令（每条指令 6 个字符；每个字 2 个字符）^。37

The main characteristic is that instructions and data are stored in the same memory device, from which any datum can be retrieved as quickly as any other. This concept arose from considering that the processing unit of a computer should not have to sit idle awaiting delivery of the next instruction. Besides that, the ratio of instructions to data usually varies for each problem, so it would not make sense to dedicate separate, expensive storage devices to each. This design implies that one may treat a coded instruction as a piece of data and perform an operation on it, thus changing it into another instruction, but that was not fully understood at first. To give a sense of how this was first implemented, the UNIVAC main store could hold up to 1,000 “words,” which could either be numbers (11 digits plus sign), characters (12 characters per word), or instructions (6 characters per instruction; 2 in each word).³⁷

最后，冯·诺依曼计算机的基本循环是将指令从存储器传输到处理器，解码该指令，并使用从同一存储器检索到的数据或处理器中已经存在的数据执行该指令。一旦处理器执行了一条指令，它就会从内存中的下一个位置获取、解码并执行另一条指令，除非另有指示。拥有快速存储设备意味着处理器可以在必要时快速分支到另一个指令流。除非遇到明确的分支指令，否则存储在内存中的指令流是顺序和线性的。38^这种获取然后执行线性指令流的概念是最持久的；即使是声称非冯·诺依曼的计算机设计通常也保留了单处理器机器的获取-解码-执行心跳。39^正如艾伦·佩利斯曾经说过的那样，“有时我认为计算领域的唯一通用方法是获取-执行周期。” ⁴⁰ UNIVAC 可以执行这个序列并在大约半毫秒内将两个数字相加。

Finally, the basic cycle of a von Neumann computer is to transfer an instruction from the store to the processor, decode that instruction, and execute it, using data retrieved from that same store or already present in the processor. Once the processor executed an instruction, it fetched, decoded, and executed another, from the very next position in memory unless directed elsewhere. Having a fast storage device meant that the processor could branch to another stream of instructions quickly whenever it was necessary. Except when explicit branch instructions are encountered, the flow through the instructions stored in the memory was sequential and linear.³⁸ This concept, of fetching and then executing a linear stream of instructions, is the most lasting of all; even computer designs that purport to be non–von Neumann typically retain the fetch-decode-execute heartbeat of a single-processor machine.³⁹ As Alan Perlis once remarked, “Sometimes I think the only universal in the computing field is the fetch-execute cycle.”⁴⁰ The UNIVAC could perform this sequence and add two numbers in about half a millisecond.

自 1990 年以来，具有并行处理结构的计算机系统变得越来越普遍，并且取指-执行周期的真正替代方案已在少数有限的市场中被接受。在其他地方，冯·诺依曼架构虽然经过了很多修改，但仍占主导地位。然而，实用并行设计的出现揭示了冯·诺依曼模型的统一效应，因为它影响了过去五十年的计算机设计。

Since 1990, computer systems with parallel processing structures have become more common, and genuine alternatives to the fetch-execute cycle have been accepted in a few limited markets. Elsewhere the von Neumann architecture, though much modified, prevails. The emergence of practical parallel designs reveals, however, the unifying effect of the von Neumann model as it influenced the computer design of the past five decades.

从 ENIAC 到 UNIVAC：第一次转型⁴¹

From ENIAC to UNIVAC: First Transformation ⁴¹

UNIVAC 将解决使用穿孔卡设备或插线板控制的复杂问题的难题，它的设计者也知道这一点。ENIAC 虽然不适合解决许多问题，但需求量如此之大，以至于它从费城到阿伯丁的物理转移不得不推迟。随着战争的结束，计算射击表的紧迫性已经降低，尽管阿伯丁试验场仍然希望将机器转移到那里用于此目的。公开亮相后，大量感兴趣的各方请愿使用它。例如，莫奇利报告说，1948 年 3 月，普惠公司问他是否能在“4 月 17 日那一周”解决一个紧急问题。这让他“笑了”——到 1948 年，ENIAC 在接下来的两年里已经被预订满了！⁴²

The UNIVAC was going to cut through the Gordian knot of solving complex problems with punched card equipment or plugboard control, and its designers knew that. The ENIAC, though ill-suited for many problems, nevertheless was in such demand that its physical transfer from Philadelphia to Aberdeen had to be put off. With the end of the War there was less urgency to compute firing tables, although the Aberdeen Proving Ground still expected the machine to be moved there for that purpose. After the public unveiling, a flood of interested parties was petitioning to use it. Mauchly reported, for example, that in March of 1948 Pratt & Whitney asked him if they could run an urgent problem “the week of April 17.” That gave him a “chuckle”—by 1948 the ENIAC was already fully booked for the next two years!⁴²

鲜为人知的是，摩尔学院的团队仔细评估了后续计算机 EDVAC 的架构，考虑到它可能要解决的问题。冯·诺依曼发现，尽管 EDVAC 最初是为评估数学表达式而设计的，但其存储程序设计使其“几乎成为一台‘万能机器’”，而且它在数据排序方面比打孔卡设备更好。这是一个至关重要的观察结果，因为排序是商业工作的核心任务，而打孔卡设备已为此进行了优化^。43

What was less well known was that the Moore School team had carefully evaluated the architecture of the follow-on computer, the EDVAC, in light of the problems it might be expected to solve. Von Neumann found that although it was initially intended for evaluating mathematical expressions, the EDVAC’s stored-program design made it “very nearly an ‘all-purpose machine’ “ and that it was better than punched card equipment for sorting data. This was a crucial observation, as sorting was a central task for commercial work, and punched card equipment had been optimized for it.⁴³

然而，埃克特-莫奇利计算机公司那一小群工程师周围的环境并不乐观。许多专家对此表示怀疑。华莱士·埃克特仍然认为，对打孔卡机进行修改，而不是采用全新的昂贵设计，才能更好地满足计算需求。霍华德·艾肯无法想象“为微分方程数值解而设计的机器的基本逻辑 [可以] 与用于为百货公司开账单的机器的逻辑相吻合。” ⁴⁴埃克特和莫奇利知道情况并非如此。UNIVAC 的逻辑结构意味着它可以做这些事情，甚至更多。这种认识促使他们和他们的公司在 20 世纪 40 年代末进入商业领域，最终诞生了 UNIVAC。

Still, the climate that surrounded the small group of engineers at the Eckert–Mauchly Computer Corporation was anything but favorable. Many experts were skeptical. Wallace Eckert still felt that modifications to punched card machines, not a radically new and expensive design, would better serve computing’s needs. Howard Aiken could not imagine that “the basic logics of a machine designed for the numerical solution of differential equations [could] coincide with the logics of a machine intended to make bills for a department store.”⁴⁴ Eckert and Mauchly knew otherwise. The UNIVAC’s logical structure meant that it could do those things and more. That knowledge drove them and their company through the late 1940s to enter the commercial area, with what eventually became the UNIVAC.

他们的行动与宾夕法尼亚大学不谋而合，后者也力图将商业利益从学院中驱逐出去。宾夕法尼亚大学的管理人员没有研究型大学的愿景来支持技术，这最终导致了加州硅谷和马萨诸塞州 128 号公路等地区的发展。摩尔学院的管理人员欧文·特拉维斯要求教职员工签署一份免责声明，以防止他们从自己的发明中获得专利使用费。他不允许讨论。埃克特和莫奇利拒绝签字。他们于 1946 年 3 月 31 日辞职。45^费城-普林斯顿地区曾是计算机技术中心的有力竞争者，但从此一蹶不振。

Their drive was matched by an equal, but opposite drive by the University of Pennsylvania to banish commercial interests from the academy. Administrators at Penn did not have the vision of a research university to support technology, which led eventually to the development of areas like Silicon Valley in California and Route 128 in Massachusetts. Irwin Travis, an administrator at the Moore School, asked that members of the staff sign a release form that would prevent them from receiving patent royalties on their inventions. He brooked no discussion. Eckert and Mauchly refused to sign. They resigned on March 31, 1946.⁴⁵ The Philadelphia-Princeton region, once a contender for the title of center for computing technology, never recovered.

埃克特和莫奇利本可以在其他大学或 IBM 找到工作，但他们选择了创办自己的公司的冒险之路。1946 年，他们成立了一家合伙企业，即电子控制公司；1948 年 12 月，他们合并为埃克特-莫奇利计算机公司。除了设计和制造通用计算机及其相关磁带驱动器、内存单元和输入输出设备的工程问题外，更大的问题就是筹集资金。美国国家标准局最初令人鼓舞；埃克特和莫奇利通过它与美国人口普查局进行了认真的讨论。（人口普查局不得签订仍在开发中的机器的合同，因此必须让国家标准局作为中间人。）人口普查局通常不被认为是技术精湛的机构，但正如它在 1890 年与赫尔曼·霍勒瑞斯合作帮助开创了现代数据处理一样，人口普查局也帮助电子计算从大学过渡到私营部门。

Eckert and Mauchly could have found work at other universities, or at IBM, but they chose instead the risky course of founding their own company. They formed a partnership, the Electronic Control Company, in 1946; in December 1948 they incorporated as the Eckert–Mauchly Computer Corporation. Added to the engineering problems of designing and building a universal computer and its associated tape drives, memory units, and input-output equipment, was the bigger problem of raising capital. The National Bureau of Standards was encouraging at first; through it Eckert and Mauchly carried out serious discussions with the U.S. Census Bureau. (Census was not allowed to contract for a machine still in development, so the NBS had to be brought in as an intermediary.) The Census Bureau is not usually considered among the technologically astute, but just as it helped inaugurate modern data processing in 1890 by working with Herman Hollerith, Census also helped make electronic computing’s transition from the university to the private sector.

但障碍仍然存在。国家统计局委托进行了一项研究，得出了对电子计算总体上，尤其是埃克特-莫奇利的保守和怀疑的结论。国家研究委员会在 1947 年进行的另一项研究得出了同样负面的结论，本章开头提到了这一点。后一项研究后来臭名昭著，因为它提出了只有少数几台计算机才能满足世界需求的说法。为了寻找资金，这家羽翼未丰的公司四处奔波：从美国博彩公司（想要一台计算机来计算赛马场的投注赔率）到诺斯罗普飞机公司（想要一种无人远程轰炸机的机载控制系统）。

Still there were roadblocks. The NBS commissioned a study, which resulted in conservative and skeptical conclusions about electronic computing in general, and Eckert–Mauchly in particular. Another study conducted by the National Research Council in 1947 produced equally negative conclusions, mentioned at the beginning of this chapter. This latter study later became infamous as the source of the statement about how only a few computers would satisfy the world’s needs. The search for funds took the fledgling company everywhere: from the American Totalisator Company, who wanted a computer to calculate betting odds at race tracks, to Northrop Aircraft, who wanted an airborne control system for an unmanned, long-range bomber.

他们疯狂地寻找资金，这让人感到沮丧。但也有好的一面：人们想要这种新机器。正如美国 Totalisator 的例子所示，除了大型军事或政府机构之外，还有许多潜在客户。

Their frantic search for capital makes for a depressing story. But it had a bright side: people wanted this new machine. And as the example of American Totalisator showed, there were many possible customers beyond the obvious ones of the large military or government agencies.

1948 年 1 月 12 日，约翰·莫奇利给埃克特-莫奇利计算机公司的员工写了一份备忘录，其中列出了他联系过的总共 22 个行业、政府机构或其他机构。他乐观地估计每个机构都是 UNIVAC 的潜在客户。46^在接下来的几年里，这家资金不足的公司将很难销售 UNIVAC。但从长远来看，莫奇利是完全正确的：这些行业以及许多其他行业都会找到令人信服的理由来购买或租赁电子数字计算机，如果不是从埃克特-莫奇利那里购买，那么就是从其他人那里购买。以下是莫奇利在备忘录中列出的一些联系人：

On January 12, 1948, John Mauchly wrote a memorandum to his staff at the Eckert–Mauchly Computer Corporation in which he listed a total of twenty-two industries, government agencies, or other institutions he had contacted. Optimistically he gauged the status of each as a potential customer for a UNIVAC.⁴⁶ In the next few years the under-capitalized company would have a great deal of trouble selling UNIVACs. But in the long run, Mauchly was exactly right: each of those industries, and many more, would find compelling reasons to purchase or lease electronic digital computers, if not from Eckert–Mauchly then from someone else. Here are some of the contacts Mauchly listed in his memo:

保诚。[埃德蒙·C·伯克利]……表示，考虑到保诚目前表示支持购买电子设备的人数，他相信订购一台 UNIVAC 不会有任何困难。

Prudential. [Edmund C. Berkeley] … . says that considering the number of persons at Prudential who have now expressed themselves in favor of obtaining electronic equipment, he believes there will be no difficulty in getting an order for one UNIVAC.

橡树岭……他们的采购订单几乎百分之百会得到陆军的批准。

Oak Ridge … . it was almost 100 percent certain that their purchase order would be approved by Army.

陆军地图服务...陆军地图服务对 UNIVAC 设备很感兴趣。

Army Map Service … . Army Map Service has taken an interest in UNIVAC equipment.

航空局……我们可能会获得一份合同。大都会保险公司有一个大问题，涉及总共 18,000,000 份保单，每周有 2,000,000 次变更。每份保单大约有 20 位信息。看来这是 UNIVAC 的自然应用……值得跟进。

Bureau of Aeronautics … . we could possibly obtain a contract. The Metropolitan Insurance Company has a large problem involving a total file of 18,000,000 policies with 2,000,000 changes per week. There are about twenty digits of information for each policy. It appears that this is a natural application for the UNIVAC …. it would be worthwhile to follow it up.

加尔各答总统学院。马哈拉诺比斯教授……一旦我们能够确定明确的条款，就急切地想要签订 UNIVAC 合同。

Presidency College, Calcutta. Professor Mahalanobis … was anxious to contract for a UNIVAC as soon as we were in a position to make definite terms.

飞机公司。许多飞机公司前景看好……毫无疑问，这些公司可以使用 UNIVAC 设备。我们曾与休斯飞机公司、格伦·L·马丁公司、联合飞机公司和北美航空公司有过短暂接触，并被告知格鲁曼公司进行了一些相当复杂的计算。

Aircraft Companies. A number of aircraft companies are good prospects … . There is no doubt that such companies could use UNIVAC equipment. We have had brief contact with Hughes Aircraft, Glen L. Martin, United Aircraft, North American Aviation, and have been told that Grumman goes in for some rather fancy calculations.

信息时代已经到来。

The Information Age had dawned.

尤尼维克

UNIVAC

我很高兴历史承认第一个发明某种东西的人，但我更关心第一个使它成功的人。

I am pleased that history recognizes the first to invent something, but I am more concerned with the first person to make it work.

—格蕾丝·霍珀⁴⁷

— Grace Hopper ⁴⁷

1951 年 3 月 31 日，雷明顿兰德公司的埃克特-莫奇利分部将第一台 UNIVAC 移交给美国人口普查局。6 月，该分部在费城里奇大街 3747 号的简陋工厂举行了正式的落成典礼。从此，美国开始了大规模存储程序计算机的商业销售时代。48^然而，这一事件并没有表面上看起来那么具有里程碑意义。第一台 UNIVAC 一直留在工厂，直到 1952 年 12 月底才被运往华盛顿。埃克特和莫奇利需要它留在那里：作为他们希望大批量销售的机器的唯一工作模型，他们想向其他潜在客户展示它。49^在经过艰苦的努力完成和调试机器之后，他们对拆卸、移动和重新设置它感到不安。第一台出厂并安装在客户场所的 UNIVAC 是序列号为 2 的机器，于 1952 年 6 月安装在五角大楼供美国空军使用。50^到1954 年，大约有 20 台 UNIVAC 被制造和销售，一套完整的系统售价约为一百万美元。51表 1.1^列出了 1951 年至 1954 年 UNIVAC 的安装情况。

On March 31, 1951, the Eckert–Mauchly Division of Remington Rand turned over the first UNIVAC to the U.S. Census Bureau. A formal dedication ceremony was held in June at the Division’s modest factory in at 3747 Ridge Avenue in Philadelphia. Thus began the era of commercial sales of large-scale stored-program computers in the United States.⁴⁸ The event was, however, less of a milestone than it appeared. That first UNIVAC remained at the plant until late December 1952, when it was shipped to Washington. Eckert and Mauchly needed it there: As the only working model of a machine they hoped to sell in quantity, they wanted to show it to other potential customers.⁴⁹ And after having gone through heroic efforts to complete and debug the machine, they were apprehensive about dismantling it, moving it, and setting it up again. The first UNIVAC to leave the factory and be installed on a customer’s premises was serial #2, installed at the Pentagon for the U.S. Air Force in June 1952.⁵⁰ By 1954 about twenty were built and sold, at prices on the order of a million dollars for a complete system.⁵¹ Table 1.1 lists UNIVAC installations from 1951 through 1954.

J. Presper Eckert 和 John Mauchly 在十几名技术人员的帮助下，设计并制造了 UNIVAC（图 1.3）。他们设计的机器使用四个二进制数字（位）来编码每个十进制数字。在其中央处理器中，四个通用累加器执行算术运算。一个字长 45 位；每个字可以表示 11 个十进制数字加一个符号，或者两个指令。UNIVAC 的时钟频率为 2.25 MHz，每秒可执行大约 465 次乘法。这与 ENIAC 的乘法速度大致相同；但是 UNIVAC 的磁带系统和存储程序架构使其整体速度更快。“延迟线”将 1,000 个字作为声音脉冲存储在水银管中，而磁带单元则可在半英寸金属带卷上存储多达一百万个字符。

J. Presper Eckert and John Mauchly, with the help of about a dozen technical employees, designed and built the UNIVAC (figure 1.3). They designed a machine that used four binary digits (bits) to code each decimal digit. In its central processor, four general-purpose accumulators carried out arithmetic. A word was 45 bits long; each word could represent 11 decimal digits plus a sign, or two instructions. The UNIVAC’s clock ran at 2.25 MHz, and it could perform about 465 multiplications per second. That was about the same as the ENIAC’s multiplication speed; but the UNIVAC’s tape system and stored-program architecture made it a much faster machine overall. “Delay lines” stored 1,000 words as acoustic pulses in tubes of mercury, while magnetic tape units stored up to one million characters on reels of half-inch metal tape.

表 1.1

UNIVAC 安装情况，1951–1954 年

Table 1.1

UNIVAC installations, 1951–1954

日期	顾客
1951 年夏	美国人口普查局
1952 年底	美国空军、五角大楼
1952 年底	美国陆军地图服务
1953 年秋季	美国原子能委员会，纽约州纽约市（纽约大学）
1953 年秋季	美国原子能委员会，加利福尼亚州利弗莫尔
1953 年秋季	大卫·泰勒模型盆地，马里兰州卡德罗克
1954	雷明顿·兰德（纽约州纽约市）
1954	通用电气，肯塔基州路易斯维尔
1954	纽约大都会人寿保险公司
1954	俄亥俄州代顿市赖特-帕特森空军基地
1954	美国钢铁公司，宾夕法尼亚州匹兹堡
1954	杜邦公司，特拉华州威尔明顿
1954	美国钢铁公司，印第安纳州加里
1954	俄亥俄州斯普林菲尔德富兰克林人寿保险公司
1954	宾夕法尼亚州匹兹堡西屋电气
1954	太平洋共同人寿保险公司，加利福尼亚州洛杉矶
1954	Sylvania Electric，纽约州纽约市
1954	纽约联合爱迪生公司
1954	纽约联合爱迪生公司

注意：此列表是根据各种来源汇编而成的，不包括已完成但仍留在雷明顿兰德的一两台 UNIVAC。在某些情况下，日期是近似的。根据对“安装”的解释，此处列出的顺序可能略有不同。UNIVAC 最后一次安装是在 1958 年末或 1959 年初。

Note: This list is compiled from a variety of sources and does not include one or two UNIVACs that were completed but remained with Remington Rand. In some cases the dates are approximate. Depending on how one interprets “installation,” the order listed here may be slightly different. UNIVACs were last installed in late 1958 or early 1959.

UNIVAC 坚固可靠。真空管故障是所有早期系统的祸根，但该故障率被控制在合理的低水平，以确保机器在实际的日常工作中仍然有用。大都会人寿保险公司的客户收集的统计数据显示，中央处理器的可用时间为 81%，与当代的真空管机器相比，这是一个非常高的数字。52 人口普查局^表示：“我们从未遇到过错误的解决方案，我们确信这是由内部计算机错误导致的。” ⁵³该机器的设计反映了 Eckert 的理念，即在真空管电路上保守负载，并具有足够的冗余度，以确保可靠运行。它的中央处理器包含 5,000 多个真空管，安装在排列在 10 英尺乘 14 英尺矩形内的机柜中。这个矩形里面是汞延迟线槽。

The UNIVAC was rugged and reliable. Vacuum tube failures, the bane of all early systems, were kept to a reasonably low rate to ensure that the machine would remain useful for practical, day-to-day work. Statistics gathered by one customer, Metropolitan Life Insurance Company, showed the central processor was available 81percent of the time, a very high figure compared to contemporary vacuum-tube machines.⁵² The Census Bureau said, “We never encountered an incorrect solution to a problem which we were sure resulted from an internal computer error.”⁵³ The machine’s design reflected Eckert’s philosophy of conservative loads on the vacuum tube circuits, plus enough redundancy, to ensure reliable operation. Its central processor contained over 5,000 tubes, installed in cabinets that were ranged in a 10-foot by 14-foot rectangle. Inside this rectangle were the mercury delay-line tanks.

图 1.3

Grace Murray Hopper 和同事坐在 UNIVAC 控制台前，约 1960 年。控制面板两侧均可见 UNIVAC 磁带卷。（来源：史密森学会照片编号 83-14878，Grace Murray Hopper 捐赠。）

Figure 1.3

Grace Murray Hopper and colleagues seated at a UNIVAC console, ca. 1960. Reels of UNIVAC tape are visible on both sides of the control panel. (Source : Smithsonian Institution photo #83-14878, gift of Grace Murray Hopper.)

许多后来变得很常见的设计特点都首先出现在 UNIVAC 中：其中包括字母数字和数字处理、大量使用额外位进行检查、用于大容量存储的磁带，以及称为“缓冲器”的电路，允许在快速延迟线和慢速磁带存储单元之间进行高速传输^。54

Many design features that later became commonplace first appeared in the UNIVAC: among them were alphanumeric as well as numeric processing, an extensive use of extra bits for checking, magnetic tapes for bulk memory, and circuits called “buffers” that allowed high-speed transfers between the fast delay line and slow tape storage units.⁵⁴

UNIVAC 的使用

The UNIVAC in Use

许多 UNIVAC 客户都是私营企业，而不是军事或国防机构。在购买 UNIVAC 的国防机构中，许多都是为了库存、物流和其他应用，这些应用在很多方面与企业客户购买机器的目的相似。简而言之，与 IBM 701（下文将讨论）相比，UNIVAC 开创了大型计算机时代，用于现在所谓的“数据处理”应用。

A number of UNIVAC customers were private corporations, not military or defense agencies. And of those defense agencies that purchased UNIVACs, many did so for inventory, logistics, and other applications that in many ways were similar to what business customers bought the machine for. In short, and in contrast to the IBM 701 (discussed next), the UNIVAC inaugurated the era of large computers for what is now called “data processing” applications.

对于大多数客户来说，UNIVAC 的革命性并不在于它的存储程序设计，甚至不在于它的电子处理器。而是用磁带代替穿孔卡。对他们来说，这台机器的“自动”特性在于它能够扫描一卷磁带，找到正确的记录或记录集，在其中执行一些处理，并将结果再次返回到磁带。在穿孔卡安装中，这些任务是由必须将大量卡片从一台穿孔卡机带到另一台穿孔卡机的人执行的。这使得穿孔卡处理劳动密集型。UNIVAC 的公开描述几乎总是将其称为“磁带”机。对于通用电气来说，“计算速度可能只是第三重要的因素。” ⁵⁵在某种程度上，客户将 UNIVAC 视为“电子大脑”，是因为它“知道”在磁带上哪里可以找到所需的数据，可以将磁带卷绕或倒回到该位置，并且可以自动提取（或记录）数据。客户将 UNIVAC 视为信息处理系统，而不是计算器。因此，它不仅取代了现有的计算机，还取代了操作计算机的人。

For most customers, what was revolutionary about the UNIVAC was not so much its stored-program design or even its electronic processor. It was the use of tape in place of punched cards. To them, the “Automatic” nature of the machine lay in its ability to scan through a reel of tape, find the correct record or set of records, perform some process in it, and return the results again to tape. In a punched card installation, these tasks were performed by people who had to carry large decks of cards from one punched card machine to another. That made punched card processing labor-intensive. Published descriptions of the UNIVAC nearly always referred to it as a “tape” machine. For General Electric, “the speed of computing is perhaps of tertiary importance only.”⁵⁵ To the extent that its customers perceived the UNIVAC as an “electronic brain,” it was because it “knew” where to find the desired data on a tape, could wind or rewind a tape to that place, and could extract (or record) data automatically. Customers regarded the UNIVAC as an information processing system, not a calculator. As such, it replaced not only existing calculating machines, but also the people who tended them.

人口普查局是这家新兴计算机公司成立的关键推动者，该局希望使用 UNIVAC 来统计 1950 年的人口普查数据。然而，当它在 1951 年收到这台机器时，大部分工作已经转移到打孔卡机上进行处理。事实上，美国空军和原子能委员会征用了生产线上的第一台机器 UNIVAC 1，用于解决联邦政府认为更紧急的问题，而人口普查局不得不暂时搁置一边^。56

The Census Bureau, which had been pivotal in getting the fledgling computer company going, hoped to use the UNIVAC for tabulating the 1950 Census. By the time it received its machine in 1951, however, much of the work had already been put on punched card machines for processing. In fact, the Census Bureau had to step aside while the U.S. Air Force and the Atomic Energy Commission commandeered the first machine off the production line, UNIVAC 1, for problems deemed more urgent by the federal government.⁵⁶

尽管如此，UNIVAC 1 还是被用于制作阿拉巴马州、爱荷华州、路易斯安那州和弗吉尼亚州第二系列人口表的一部分。这涉及将个人分为几百个组之一，进一步按地理位置分组，并准备表格，显示每个地区每个组中的人数。这项操作的数据最初打在 1100 万张卡片上（每人一张），然后被转移到磁带上，供 UNIVAC 处理。57^这台机器还用于制表另一个涉及约 500 万户家庭的人口子集。每个问题都需要几个月的时间才能完成。

Nevertheless, UNIVAC 1 was used for the production of part of the Second Series Population Tables for the states of Alabama, Iowa, Louisiana, and Virginia. This involved classifying individuals into one of several hundred groups, further grouping them by geographic location, and preparing tables showing the number of persons in each group for each local area. The data for this operation, initially punched onto eleven million cards (one for each person), was transferred to tape for processing by the UNIVAC.⁵⁷ The machine was also used for tabulating another subset of population involving about five million households. Each problem took several months to complete.

UNIVAC 2 安装在五角大楼，供空军审计长使用，旨在用于 SCOOP（最优问题的科学计算）项目，该项目源于战时对将战争物资和人员运送到大西洋彼岸的担忧。战后，新成立的空军面临着数学上类似的问题，即维护和供应遍布全球的空军基地。SCOOP 项目在发现线性规划方面发挥了关键作用，线性规划是现代应用数学的基石^。58

UNIVAC 2, installed at the Pentagon for the Air Comptroller, was intended for use in Project SCOOP (Scientific Computation of Optimum Problems), which grew out of wartime concerns with getting war materials and men across the Atlantic. Following the War, the newly created Air Force was faced with a mathematically similar problem in maintaining and supplying air bases scattered across the globe. Project SCOOP played a key role in the discovery of Linear Programming, a cornerstone of modern applied mathematics.⁵⁸

空军曾为 SCOOP 资助建造了一台名为 SEAC（标准东方自动计算机）的计算机，但该机器的输入/输出功能有限，因此不太适合解决此问题。安装后不久，UNIVAC 2 便开始全天候为 SCOOP 工作。59 尽管^UNIVAC在许多方面都优于 SEAC，但它的输出机制也很慢，这妨碍了它在 SCOOP 中的使用。UNIVAC 的 UNIPRINTER 基于标准的雷明顿兰德电动打字机，其打印速度与这种机器相当，约为每秒 10 个字符，这对于 UNIVAC 的销售数据处理应用来说太慢了。1954 年，雷明顿兰德通过推出 UNIVAC 高速打印机解决了这个问题，该打印机一次可打印整整 130 个字符^。60

It was for SCOOP that the Air Force had helped fund construction of a computer called SEAC (Standards Eastern Automatic Computer), but that machine’s limited Input/Output facilities made it less than ideal for this problem. Soon after its installation, UNIVAC 2 was put to work on SCOOP around the clock.⁵⁹ Although the UNIVAC was superior to the SEAC in many ways, it, too, suffered from a slow output mechanism, which hampered its use for SCOOP. The UNIVAC’s UNIPRINTER was based around a standard Remington Rand electric typewriter, and it printed at a rate commensurate with such a machine, about ten characters per second, which was too slow for the data processing applications the UNIVAC was being sold for. In 1954 Remington Rand addressed the problem by introducing the UNIVAC High Speed Printer, which printed a full 130-character line at one time.⁶⁰

1954 年，位于俄亥俄州赖特-帕特森空军基地的空军物资司令部安装了一台 UNIVAC，执行了类似的任务。它的首要任务之一是计算“1956 财年机载设备备件的完整预算，涉及约 50 万件”。61^空军指出，这台机器在一天内就完成了这项工作，取代了一组打孔卡设备。

The UNIVAC installed in 1954 at Air Force’s Air Material Command at Wright-Patterson AFB in Ohio performed similar tasks. One of its first jobs was to calculate “the complete Fiscal 1956 Budget estimate for airborne equipment spare parts, involving approximately 500,000 items.”⁶¹ The Air Force noted that the machine did the job in one day, replacing a battery of punched card equipment.

一些 UNIVAC 计算机执行机密武器工作，其精神与之前独一无二的计算机相似。1953 年 4 月安装在劳伦斯利弗莫尔实验室的 UNIVAC 5 就是其中之一。但即使是这台机器也至少进行了一次不用于武器设计的计算。1952 年 11 月，在运往加利福尼亚之前，雷明顿兰德公司用它预测了艾森豪威尔在 1952 年总统大选中击败阿德莱史蒂文森的结果。这一事件在“现场”电视上播出，标志着电视进入国家政治，计算机进入公众意识。在一段短暂的时期内，“UNIVAC”一词与计算机同义，就像“Thermos”与真空瓶同义一样。当 IBM 在该业务中占据领先地位时，这种情况就结束了^。62

Some UNIVACs performed classified weapons work in the spirit of the one-of-a-kind computers that preceded them. UNIVAC 5, installed at the Lawrence Livermore Labs in April 1953, was one of those. But even that machine did at least one calculation that was not for the purpose of weapons designs. In November 1952, before it was shipped to California, Remington Rand used it to predict Eisenhower’s victory over Adlai Stevenson in the 1952 presidential election. Narrated on “live” television, the event inaugurated the intrusion of television into national politics, and of computers into the public’s consciousness. For a brief period, the word “UNIVAC” was synonymous with computer, as “Thermos” was for vacuum bottles. That ended when IBM took the lead in the business.⁶²

UNIVAC 的最后使用实例来自通用电气公司位于肯塔基州路易斯维尔郊外的家电园区。该园区于 1954 年安装，作为第一台为非政府客户提供的存储程序电子计算机而闻名（尽管 LEO 是为伦敦的 J. Lyons Catering Company 制造的，比它早了三年）。

A final example of the UNIVAC in use comes from the experience at General Electric’s Appliance Park, outside Louisville, Kentucky. This installation, in 1954, has become famous as the first of a stored-program electronic computer for a nongovernment customer (although the LEO, built for the J. Lyons Catering Company in London, predated it by three years).

在路易斯维尔的 Roddy F. Osborn 的指导下，以及在芝加哥咨询公司 Arthur Andersen & Co. 的建议下，通用电气购买了一台 UNIVAC，用于四项特定任务：工资单、材料调度和库存控制、订单服务和计费以及一般成本会计。63^这些都是平淡无奇的操作，但通用电气还希望计算机不仅仅是当时使用的打孔卡设备的替代品。对于通用电气以及美国工业而言，UNIVAC 是迈向“自动化”时代的第一步，这一变革对商业的革命性不亚于半个世纪前弗雷德里克·W·泰勒的科学管理。

Under the direction of Roddy F. Osborn at Louisville, and with the advice of the Chicago consulting firm Arthur Andersen & Co., General Electric purchased a UNIVAC for four specific tasks: payroll, material scheduling and inventory control, order service and billing, and general cost accounting.⁶³ These were prosaic operations, but GE also hoped that the computer would be more than just a replacement for the punched-card equipment in use at the time. For General Electric, and by implication for American industries, the UNIVAC was the first step into an age of “automation,” a change as revolutionary for business as Frederick W. Taylor’s Scientific Management had been a half-century earlier.

“自动化”一词于 1947 年由福特汽车公司发明，并由约翰·迪博尔德在 1952 年出版的同名书中推广。64^迪博尔德将这个词定义为将“反馈”机制应用于商业和工业实践，以计算机为主要工具。他谈到 20 世纪 50 年代是“按钮时代已经过时；按钮现在会自动按下”的时代。65罗迪^·奥斯本在描述通用电气的安装时预测，UNIVAC 将对商业产生与它已经开始在科学、工程和数学中产生的影响相同的变化。“虽然科学家和工程师已经完全清醒地利用这些非凡的工具取得进展，但企业，就像瑞普·凡·温克尔一样，却一直处于沉睡状态。通用电气安装 UNIVAC 可能是瑞普·凡·温克尔商业的第一次‘眨眼’。” ⁶⁶

The term “automation” was coined at the Ford Motor Company in 1947 and popularized by John Diebold in a 1952 book by that title.⁶⁴ Diebold defined the word as the application of “feedback” mechanisms to business and industrial practice, with the computer as the principal tool. He spoke of the 1950s as a time when “the push-button age is already obsolete; the buttons now push themselves.”⁶⁵ Describing the GE installation, Roddy Osborn predicted that the UNIVAC would effect the same kind of changes on business as it had already begun to effect in science, engineering, and mathematics. “While scientists and engineers have been wide-awake in making progress with these remarkable tools, business, like Rip Van Winkle, has been asleep. GE’s installation of a UNIVAC may be Rip Van Business’s first ‘blink.’ ”⁶⁶

对于通用电气的员工来说，这些关于“电子大脑”和“自动化”的说法是一把双刃剑。路易斯维尔工厂的设计和建造都力求做到尽可能现代化和先进；这也是将工厂设在肯塔基州而不是马萨诸塞州或纽约州的动机，因为在马萨诸塞州或纽约州，传统方法（和工会）占主导地位。与此同时，通用电气需要向股东保证，它不会仅仅因为“长发”学者（不关心利润）想要这样做，就着手购买奇特、易碎且昂贵的设备。

To people at General Electric, these accounts of “electronic brains” and “automation” were a double-edged sword. The Louisville plant was conceived of and built to be as modern and sophisticated as GE could make it; that was the motivation to locate it in Kentucky rather than Massachusetts or New York, where traditional methods (and labor unions) held sway. At the same time, GE needed to assure its stockholders that it was not embarking on a wild scheme of purchasing exotic, fragile, and expensive equipment just because “longhair” academics— with no concern for profits—wanted it to.

因此，通用电气必须强调四项已经由打孔卡设备完成的平凡工作，以证明 UNIVAC 的合理性。一旦这些工作成为常规工作，其他更先进的工作将交给机器。虽然使用小型计算机可以自动完成这四项任务，但通用电气选择 UNIVAC 是因为它预见到有一天会完成更复杂的工作。这些任务将涉及长期规划、基于人口统计数据的市场预测、改进生产流程以减少库存和运输延误，以及需要更积极地使用公司信息的类似工作。67^更先进的应用程序要等到现有的“面包和黄油”工作计算机化达到“盈亏平衡点……足以让管理层相信计算机系统可以通过直接节省美元（从工资单上删除人员）来收回成本，而无需等待更迷人的应用程序的‘拥堵’。” ⁶⁸

Thus, GE had to emphasize the four mundane jobs, already being done by punched card equipment, to justify the UNIVAC. Once these jobs became routine, other, more advanced jobs would be given to the machine. Although automating those four tasks could have been done with a smaller computer, GE chose a UNIVAC in anticipation of the day when more sophisticated work would be done. These tasks would involve long-range planning, market forecasting based on demographic data, revamping production processes to reduce inventories and shipping delays, and similar jobs requiring a more ambitious use of corporate information.⁶⁷ The more advanced applications would not commence until after the existing computerization of “bread and butter” work reached a “break even point … enough to convince management that a computer system can pay for itself in terms of direct dollar savings (people off the payroll) without waiting for the ‘jam’ of more glamorous applications.”⁶⁸

事实上，对 UNIVAC 效益的分析几乎完全是针对其取代领薪职员的能力以及办公空间、家具和福利等间接成本。然而，在奥斯本为《哈佛商业评论》撰写的文章结尾处，编辑们附上了西奥多·卡洛当年出版的《工作社会学》中的一段引文。这段引文开头是这样的：

Indeed, the analysis of the UNIVACs benefits was almost entirely cast in terms of its ability to replace salaried clerks and their overhead costs of office space, furnishings, and benefits. Yet at the end of Osborn’s essay for the Harvard Business Review, the editors appended a quotation from Theodore Callow’s The Sociology of Work, published that year. That quotation began:

自动化生产的乌托邦本质上是可行的。事实上，当今美国的贫困已经意味着缺乏地位象征，而不是饥饿和身体痛苦，与前几代人的预算经验或其他大陆上大多数人的当代经验相比，这种状况令人惊叹地有利^。69

The Utopia of automatic production is inherently plausible. Indeed, the situation of the United States today, in which poverty has come to mean the absence of status symbols rather than hunger and physical misery, is awesomely favorable when measured against the budgetary experience of previous generations or the contemporary experience of most of the people living on the other continents.⁶⁹

这不会是计算机最后一次被视为带来数字乌托邦的机器。

It would not be the last time that the computer would be seen as the machine that would bring on a digital Utopia.

1954 年 10 月 15 日星期五，通用电气 UNIVAC 首次为电器园区的员工制作工资支票。70^打孔卡机多年来一直在做这项工作，但对于将数据记录为磁带卷上不可见磁点的电子数字计算机来说，这是一个里程碑。工资单必须正确按时发放。通用电气已经彻底演练了这次转换，他们已经与雷明顿兰德公司商定，如果他们的机器发生故障并导致支票延迟发放，他们可以将磁带带到另一个 UNIVAC 客户那里，在那里完成这项工作。71^在接下来的一年里，他们必须至少行使一次这个选择权。有几次支票是在最后一刻打印出来的，在最初的几个月里，使用 UNIVAC 完成这项工作所花的时间通常比使用打孔卡设备所花的时间多得多。工资单从未延迟发放。

On Friday, October 15, 1954, the GE UNIVAC first produced payroll checks for the Appliance Park employees.⁷⁰ Punched-card machines had been doing that job for years, but for an electronic digital computer, which recorded data as invisible magnetic spots on reels of tape, it was a milestone. Payroll must be done right, and on time. GE had rehearsed the changeover thoroughly, and they had arranged with Remington Rand that if their machine broke down and threatened to make the checks late, they could bring their tapes to another UNIVAC customer and run the job there.⁷¹ Over the course of the next year they had to exercise this option at least once. There were several instances where the checks were printed at the last possible minute, and in the early months it was common to spend much more time doing the job with UNIVAC than had been spent with punched card equipment. No payrolls were late.

IBM 的回应

IBM’s Response

在 UNIVAC 发布时，IBM 尚未完全致力于电子计算，而是大力推销其穿孔卡计算器和制表机系列。但在看到竞争威胁后，它推出了几台机器作为回应：两台与 UNIVAC 相当；另一台则更普通。

At the time of the UNIVAC’s announcement, IBM was not fully committed to electronic computation and was vigorously marketing its line of punched card calculators and tabulators. But after seeing the competitive threat, it responded with several machines: two were on a par with the UNIVAC; another was more modest.

1952 年 5 月，IBM 发布了 701，这是一款与 UNIVAC 同级别的存储程序计算机。虽然不是完全复制品，但它的设计与约翰·冯·诺依曼在普林斯顿高等研究院建造的计算机非常相似。这意味着它使用一个内存设备，可以一次检索一个字的所有数字，而不是 UNIVAC 的延迟线，每次检索一位。从那年 1 月开始，IBM 聘请了约翰·冯·诺依曼作为顾问；与高等研究院计算机本身一样，冯·诺依曼没有参与 701 的详细设计。（IBM 工程师 Jerrier Haddad 和 Nat Rochester 负责该项目。）第一台机器于 12 月安装在 IBM 纽约办事处，第一批货物于 1953 年初运往洛斯阿拉莫斯的核武器实验室^。72

In May 1952, IBM announced the 701, a stored-program computer in the same class as the UNIVAC. Although not an exact copy, its design closely followed that of the computer that John von Neumann was having built at the Institute for Advanced Study at Princeton. That meant it used a memory device that retrieved all the digits of a word at once, rather than the UNIVAC’s delay lines that retrieved bits one at a time. Beginning in January of that year, IBM had hired John von Neumann as a consultant; as with the Institute for Advanced Study computer itself, von Neumann was not involved with the detailed design of the 701. (IBM engineers Jerrier Haddad and Nat Rochester were in charge of the project.) The first unit was installed at IBM’s offices in New York in December, with the first shipment outside IBM to the nuclear weapons laboratory at Los Alamos in early 1953.⁷²

IBM 将 701 称为“电子数据处理机”，该术语（由 James Birkenstock 创造）与“电子会计机”非常契合，IBM 用后者来描述其新的穿孔卡设备系列。IBM 刻意避免使用“计算机”一词，因为它认为该词与 UNIVAC 以及看起来与业务无关的奇特战时项目密切相关。

IBM called the 701 an “electronic data processing machine,” a term (coined by James Birkenstock) that fit well with “Electric Accounting Machine,” which IBM was using to describe its new line of punched card equipment. IBM deliberately avoided the word “computer,” which it felt was closely identified with the UNIVAC and with exotic wartime projects that appeared to have little relevance to business.

对于主存储，701 使用了 IBM 设计的 3 英寸直径真空管，类似于电视机中使用的真空管。（它们被称为“威廉姆斯管”，以英国发明家 FC Williams 命名。）虽然它们比其他当代计算机中的管更可靠，但它们的不可靠性是系统中的一个薄弱环节。有一个故事讲述了一台 701 在向媒体发布时表现异常，尽管在仪式前已经进行了彻底检查。摄影师的闪光灯“使”威廉姆斯管“失明”，导致它们丢失数据。另一个说法是，由于内存的平均故障间隔时间 (MTBF) 只有 20 分钟，因此必须不断将数据交换到磁鼓中以防止丢失^。73

For main storage, the 701 used IBM-designed 3-inch diameter vacuum tubes similar to those used in television sets. (They were called “Williams tubes” after their British inventor, F. C. Williams.) Although they were more reliable than those in other contemporary computers, their unreliability was a weak link in the system. One story tells of a 701 behaving erratically at its unveiling to the press despite having been checked out thoroughly before the ceremony. The photographers’ flash bulbs were “blinding” the Williams tubes, causing them to lose data. Another account said that because the memory’s Mean Time Between Failure (MTBF) was only twenty minutes, data had to be constantly swapped to a drum to prevent loss.⁷³

每个管子设计为可容纳 1,024 位。因此，72 个管子组成的阵列可以容纳 2,048 个 36 位字，并且每次传输一个字，方法是从 36 个管子中各读取一个位。74^涂有磁性氧化物的塑料带用作大容量存储器，鼓用作中间存储器。该处理器每秒可执行约 2,000 次乘法，比 UNIVAC 快四倍左右。

Each tube was designed to hold 1,024 bits. An array of 72 tubes could thus hold 2,048 36-bit words, and transfer a word at a time by reading one bit from each of 36 tubes.⁷⁴ Plastic tape coated with magnetic oxide was used for bulk memory, with a drum for intermediate storage. The processor could perform about 2,000 multiplications/second, which was about four times faster than the UNIVAC.

在 IBM 内部，701 被称为国防计算器，以适应其预期市场。据 IBM 一位高管称，这个名字还有助于“缓解一些内部反对意见，因为它可以被视为一个特殊项目（如 IBM 在二战期间制造的轰炸瞄准器、步枪等），并不打算威胁 IBM 的主要产品线。” ⁷⁵正如这种看法所言，安装的 19 种型号中几乎所有型号都是美国国防部或军事航空航天公司。76^最初的租赁费为每月 15,000 美元；IBM 并没有直接出售这些机器。如果我们假设 701 是一台像 UNIVAC 一样价值百万美元的机器，那么租赁价格似乎很低；当然，IBM 不可能在机器成为可行产品的几年内收回成本。

Within IBM, the 701 had been known as the Defense Calculator, after its perceived market. According to an IBM executive, the name also helped “ease some of the internal opposition to it since it could be viewed as a special project (like the bomb sights, rifles, etc., IBM had built during World War II) that was not intended to threaten IBM’s main product line.”⁷⁵ True to that perception, nearly all of the 19 models installed were to U.S. Defense Department or military aerospace firms.⁷⁶ Initial rental fees were $15,000 a month; IBM did not sell the machines outright. If we assume the 701 was a million-dollar machine like the UNIVAC, the rental price seems low; certainly IBM could not have recouped its costs in the few years that the machine was a viable product.

701 客户最初使用这台机器解决许多仍属机密的问题，包括武器设计、航天器轨迹和密码分析，这些问题对中央处理器的负荷比其输入/输出设备更大。打孔卡设备一直在做一些这样的工作，但也曾用计算尺、机械计算器、模拟计算机和卡片编程计算器做过。然而，最终客户将 701 应用于 UNIVAC 所做的同类工作：军事机构的后勤、财务报告、精算报告、工资单（北美航空），甚至预测网络电视的总统选举结果。（1956 年，701 正确预测了艾森豪威尔的连任。）⁷⁷

The 701 customers initially used the machine for problems, many still classified, involving weapons design, spacecraft trajectories, and cryptanalysis, which exercised the central processor more heavily than its Input/Output facilities. Punched card equipment had been doing some of that work, but it had also been done with slide rules, mechanical calculators, analog computers, and the Card-Programmed Calculator. Eventually, however, customers applied the 701 to the same kinds of jobs the UNIVAC was doing: logistics for a military agency, financial reports, actuarial reports, payrolls (for North American Aviation), and even predicting the results of a presidential election for network television. (In 1956, the 701 correctly predicted Eisenhower’s reelection.)⁷⁷

与 UNIVAC 不同，701 的中央处理器直接控制慢速输入/输出 (I/O) 设备。所有数据传输都必须通过机器处理器中的单个寄存器，这导致需要大量使用 I/O 的任务运行缓慢。但是，701 的轻质塑料磁带可以比 UNIVAC 的金属磁带更快地启动和停止，从而加快这些操作。磁带驱动器还采用了 James Wiedenhammer 发明的巧妙的真空柱机制，使磁带能够快速启动和停止而不会撕裂。

Unlike the UNIVAC, the 701’s central processor handled control of the slow input/output (I/O) facilities directly. All transfers of data had to pass through a single register in the machine’s processor, which led to slow operation for tasks requiring heavy use of I/O. However, the 701’s lightweight plastic tape could start and stop much faster than the UNIVAC’s metal tape and thus speed up those operations. The tape drive also employed an ingenious vacuum-column mechanism, invented by James Wiedenhammer, which allowed the tape to start and stop quickly without tearing.

对于科学和工程问题，701 的不平衡 I/O 并不是一个严重的障碍。计算机设计师（1953 年为数不多）认为它是一种不雅的设计，但客户喜欢它。19 台安装足以阻止 UNIVAC 完全占领市场，并开始 IBM 向设计和制造大型电子数字计算机的公司转型^。78

For scientific and engineering problems, the 701’s unbalanced I/O was not a serious hindrance. Computer designers—the few there were in 1953—regarded it as an inelegant design, but customers liked it. The nineteen installations were enough to prevent UNIVAC from completely taking over the market and to begin IBM’s transition to a company that designed and built large-scale electronic digital computers.⁷⁸

701 成为 IBM 在市场上对 UNIVAC 的回应，但这并不是 IBM 的本意。在开始 701 之前，IBM 已经开发了一个研究项目，研究一种类似于 UNIVAC 的机器，一种称为磁带处理机或 TPM 的实验机器。它的设计于 1950 年 3 月完成。79 ^TPM在两个方面与 IBM 的穿孔卡机器截然不同。它使用磁带（像 UNIVAC），其可变长度记录取代了穿孔卡所强加的严格的 80 个字符格式。与 UNIVAC 一样，它使用十进制数字，以二进制对每个数字进行编码。

The 701 became IBM’s response to UNIVAC in the marketplace, but that had not been IBM’s intention. Before starting on the 701, IBM had developed a research project on a machine similar to the UNIVAC, an experimental machine called the Tape Processing Machine, or TPM. Its design was completed by March 1950.⁷⁹ The TPM was a radical departure from IBM’s punched card machinery in two ways. It used magnetic tape (like the UNIVAC), and its variable length record replaced the rigid 80-character format imposed by the punched card. Like the UNIVAC, it worked with decimal digits, coding each digit in binary.

IBM 选择向商业客户推销第二台基于磁带处理机的大型计算机。702 型于 1953 年 9 月发布，并于 1955 年交付。它在许多方面与 701 相似，使用大部分相同的电子电路以及 Williams Tube 存储器。到第一批 702 安装时，磁芯存储器开始用于商用机器。701 客户发现他们的机器与任何强大的通用计算机一样，也可以用于商业应用。IBM 收到了许多 702 的订单，但选择只制造和交付 14 台，其他订单由 IBM 几年后推出的另一台机器完成^。80

IBM chose to market a second large computer specifically to business customers based on the Tape Processing Machine. Model 702 was announced in September 1953 and delivered in 1955. In many ways it was similar to the 701, using most of the same electronic circuits as well as the Williams Tube storage. By the time of the first 702 installations, magnetic core memories were beginning to be used in commercial machines. And 701 customers were finding that their machine, like any powerful general-purpose computer, could be used for business applications as well. IBM received many orders for 702s, but chose to build and deliver only fourteen, with other orders filled by another machine IBM brought out a few years later.⁸⁰

工程研究助理

Engineering Research Associates

第三家公司于 20 世纪 50 年代初进入大型数字计算机制造和销售领域：工程研究协会，这是一家位于双子城的公司，起源于二战期间美国海军赞助的密码破译活动。81^海军将这项工作命名为“通信补充活动——华盛顿”（CSAW），但通常以其首字母缩写词“Seesaw”命名。它的中心位于华盛顿，位于一所女子学校的征用校园内。战后，该组织的两名成员 Howard Engstrom 和 William Norris 认为海军为战争努力聚集的人才和技能太宝贵了，不能分散，他们探索了让该组织保持团结的方法。他们决定成立一家私人公司，并在 John E. Parker 的财政援助下，于 1946 年初成立了工程研究协会有限公司。Parker 能够在圣保罗的一栋建筑中提供空间，该建筑在战争期间生产木制滑翔机（包括用于诺曼底入侵的滑翔机）。

A third firm entered the field of making and selling large digital computers in the early 1950s: Engineering Research Associates, a Twin Cities firm that had its origins in U.S. Navy-sponsored code-breaking activities during World War II.⁸¹ The Navy gave this work the name “Communications Supplementary Activity—Washington” (CSAW), but it was usually called “Seesaw” after its acronym. It was centered in Washington, on the commandeered campus of a girls school. After the War, two members of this group, Howard Engstrom and William Norris, felt that the talent and skills the Navy had assembled for the war effort were too valuable to be scattered, and they explored ways of keeping the group together. They decided to found a private company, and with financial assistance from John E. Parker, they were incorporated as Engineering Research Associates, Inc., in early 1946. Parker was able to provide space in a St. Paul building that during the war had produced wooden gliders (including those used for the Normandy invasion).

因此，在这段历史中，巧合时有发生，空荡荡的滑翔机工厂让双城进入了先进的数字计算世界。工厂寒冷多风，但 ERA 毫不费力地就找到并聘用了刚从该地区工程学校毕业的能干工程师。其中有一位 1951 年毕业于明尼苏达大学的毕业生，他去了“滑翔机工厂”，因为他听说那里可能有工作。他的名字是西摩·R·克雷。82^在后面的章节中，我们将多次遇到克雷和他的老板威廉·诺里斯。

Thus, by one of the coincidences that periodically occur in this history, the empty glider factory gave the Twin Cities an entree into the world of advanced digital computing. The factory was cold and drafty, but ERA had little trouble finding and hiring capable engineers freshly minted from the region’s engineering schools. Among them was a 1951 graduate of the University of Minnesota, who went over to “the glider factory” because he heard there might be a job there. His name was Seymour R. Cray.⁸² We will encounter Cray and his boss, William Norris, several times in later chapters.

ERA 是一家私营公司，但同时也隶属于海军，它就是从海军发展而来的。（这种安排的合理性有时会引发问题，但并不严重。）海军给它分配了一些工作或“任务”，由 ERA 执行。其中大部分都是高度机密的，与破译密码业务有关。1947 年 8 月分配的任务 13 是开发一台通用电子计算机。ERA 完成了代号为“Atlas”的机器，并要求海军批准他们开发一个可以在公开市场上销售的非机密版本。1951 年 12 月，他们宣布它为型号“1101”：二进制表示为“13” ^。83

ERA was a private company but was also captive to the Navy, from which it had sprung. (The propriety of this arrangement would on occasion cause problems, but none serious.) The Navy assigned it a number of jobs, or “tasks,” that ERA carried out. Most of these were highly classified and related to the business of breaking codes. Task 13, assigned in August 1947, was for a general-purpose electronic computer. ERA completed the machine, code-named “Atlas,” and asked the Navy to clear them for an unclassified version they could sell on the open market. In December 1951 they announced it as Model “1101”: “13” in binary notation.⁸³

正如人们对 ERA 这样的公司所预料的那样，1101 是为科学或工程客户设计的，其设计也反映了这一点。然而，在开始交付系统之前，ERA 发现自己需要的资金远远超过其创始人所能提供的资金，因此像 Eckert–Mauchly 计算机公司一样，它被雷明顿兰德公司收购。到 1952 年中期，雷明顿兰德公司可以提供不止一款，而是两款设计精良、性能强大的计算机系统，一款针对科学和工程进行了优化，另一款则用于商业用途。其继任者 1103 的安装始于 1953 年秋季。大约建造了 20 台。与 IBM 701 一样，大多数都流向了军事机构或航空航天公司。

As might be expected from a company like ERA, the 1101 was intended for scientific or engineering customers, and its design reflected that. Before it could begin delivering systems, however, ERA found itself needing much more capital than its founders could provide, and like the Eckert–Mauchly Computer Corporation, was purchased by Remington Rand. By mid-1952 Remington Rand could offer not one but two well-designed and capable computer systems, one optimized for science and engineering, the other for commercial use. Installations of the 1103, its successor, began in the fall of 1953. Around twenty were built. As with the IBM 701, most went to military agencies or aerospace companies.

1954 年，该公司向美国国家航空咨询委员会 (NACA) 交付了一台 1103，该机器使用磁芯代替威廉姆斯管存储器。这也许是磁芯首次用于商用机器。1103 使用二进制算术，字长为 36 位，每次对一个字的所有位进行操作。威廉姆斯管提供 1,024 个字的主存储器，配有 ERA 设计的磁鼓和四个磁带单元作为二级存储。84 根据^NACA的建议，ERA 修改了机器的指令集，包括“中断”功能 - 这是计算机设计中的另一个首创。（磁芯和中断将在下一章详细讨论。）这些增强功能后来作为 1103-A 型号的标准功能进行营销。85^另一家航空航天客户康维尔为 1103 开发了一种 CRT 管显示器，他们称之为 Charactron。这个 7 英寸的管子能够显示 6×6 的字符阵列，这也影响了计算机历史的进程。86^总体而言，1103 与 IBM 701 竞争激烈，尽管其 I/O 设施被认为略逊一筹。

In 1954 the company delivered an 1103 to the National Advisory Committee for Aeronautics (NACA) that employed magnetic core in place of the Williams Tube memory. This was perhaps the first use of core in a commercial machine. The 1103 used binary arithmetic, a 36-bit word length, and operated on all the bits of a word at a time. Primary memory of 1,024 words was supplied by Williams tubes, with an ERA-designed drum, and four magnetic tape units for secondary storage.⁸⁴ Following NACA’s advice, ERA modified the machine’s instruction set to include an “interrupt” facility—another first in computer design. (Core and interrupts will be discussed in detail in the next chapter.) These enhancements were later marketed as standard features of the 1103-A model.⁸⁵ Another aerospace customer, Convair, developed a CRT tube display for the 1103, which they called the Charactron. This 7-inch tube was capable of displaying a 6 × 6 array of characters, which also affected the course of computer history.⁸⁶ Overall, the 1103 competed well with the IBM 701, although its I/O facilities were judged somewhat inferior.

鼓机

The Drum Machines

20 世纪 30 年代末，JV Atanasoff 设想了一种存储设备，该设备由一个旋转鼓组成，鼓上放置了 1,600 个电容器，排列成 32 行，这可能是首次尝试制造电子数字计算机。87^他的工作影响了接下来十年的发展，尽管后来的人最终没有采用他的方法。在接下来的几年里，一些人继续研究旋转磁性设备用于数据存储的想法，例如 Perry O. Crawford，他在麻省理工学院的硕士论文中描述了这种设备^。88

In the late 1930s, in what may have been the first attempt to build an electronic digital computer, J. V. Atanasoff conceived of a memory device consisting of a rotating drum on which 1,600 capacitors were placed, arrayed in 32 rows.⁸⁷ His work influenced the developments of the next decade, although those who followed him did not ultimately adopt his method. In the following years several people continued to work on the idea of rotating magnetic devices for data storage, for example, Perry O. Crawford, who described such a device in his master’s thesis at MIT.⁸⁸

战后，磁鼓成为一种可靠、坚固、廉价但速度较慢的存储设备。设计师借鉴了美国和德国在战时对磁记录的研究，重新发现并完善了磁鼓，这次使用的是磁性技术而非电容技术。

After the War, the drum emerged as a reliable, rugged, inexpensive, but slow memory device. Drawing on wartime research on magnetic recording in both the United States and Germany, designers rediscovered and perfected the drum, this time using magnetic rather than capacitive techniques.

这项工作的领导者是工程研究协会 (ERA)。在被指派“任务 13”之前，他们被要求研究可用的内存技术。到 1947 年，他们使用粘有氧化物涂层纸的磁鼓在记录速度和密度方面取得了一些重大进展 (图 1.4 )。89^两年内，ERA 制造了直径从 4.3 到 34 英寸不等的磁鼓，容量高达 200 万位，或 65,000 个 30 位字。访问时间范围为 8 到 64 毫秒。90 ^ERA在 1101 中使用了磁鼓；他们还向其他人宣传出售该技术。

The leader in this effort was Engineering Research Associates. Before they were assigned “Task 13,” they were asked to research available memory technologies. By 1947 they had made some significant advances in recording speeds and densities, using a drum on which they had glued oxide-coated paper (figure 1.4).⁸⁹ Within two years ERA was building drums that ranged from 4.3 to 34 inches in diameter, with capacities of up to two million bits, or 65,000 30-bit words. Access time ranged from 8 to 64 milliseconds.⁹⁰ ERA used drums in the 1101; they also advertised the technology for sale to others.

CRC 102A

最早利用磁鼓的公司之一是加利福尼亚州霍桑的计算机研究公司。该公司由诺斯罗普飞机公司的前雇员创立，诺斯罗普飞机公司制造了上述卡片编程计算器。1953 年，他们开始销售 CRC-102A，这是为空军制造的计算机 CADAC 的生产版本。这是一款基于磁鼓存储器的存储程序通用计算机。102A 的设计简单，使用二进制算法，但 1954 年推出了十进制版本 (CRC 102D)。91 在^一些已发表的描述中，工程师将其设计描述为直接基于布尔代数语句得出的逻辑状态。这种所谓的西海岸设计被视为与埃克特和莫奇利的设计不同，后者不是从逻辑状态的角度思考，而是从通过机器各个部件的电流脉冲的角度思考。随着计算机工程的成熟，两种设计方法的元素融合在一起，这种区别最终消失了。⁹²

One of the first to take advantage of magnetic drums was was Computer Research Corporation of Hawthorne, California. This company was founded by former employees of Northrop Aircraft Company, the company that had built the Card-Programmed Calculator described above. In 1953 they began selling the CRC-102A, a production version of a computer called CADAC that had been built for the Air Force. It was a stored-program, general-purpose computer based on a drum memory. The 102A had a simple design, using binary arithmetic, but a decimal version (CRC 102D) was offered in 1954.⁹¹ In some of the published descriptions, engineers describe its design as based directly on logic states derived from statements of Boolean algebra. This so-called West Coast design was seen as distinct from the designs of Eckert and Mauchly, who thought in terms not of logic states, but of current pulses gated through various parts of a machine. As computer engineering matured, elements of both design approaches merged, and the distinction eventually vanished.⁹²

图 1.4

ERA 的磁鼓存储单元广告。（来源：《电子杂志》 [1953 年 4 月]：397。）

Figure 1.4

Advertisement for magnetic drum memory units, from ERA. (Source : Electronics Magazine [April 1953]: 397.)

102A 的磁鼓存储器存储了 1,024 个 42 位字；平均访问时间为 12.5 毫秒。磁带系统存储了另外 100,000 个字。主要的输入和输出设备是 Flexowriter，这是一种类似打字机的设备，可以存储或读取纸带上的按键。它的运行速度与普通电动打字机差不多，它是从普通电动打字机衍生而来的。为了保持其航空航天根源，Computer Research Corporation 还提供了一个转换器，用于将图形或其他模拟数据输入机器。93^还可以将 IBM 读卡器或打孔机连接到计算机。计算机的运行速度估计为每秒约 11 次乘法。94 ^102A是一款平衡性良好的计算机，销量适中。1954 年，国家收银机公司收购了 CRC，102 成为 NCR 进入计算机业务的基础^。95

The 102A’s drum memory stored 1,024 42-bit words; average access time was 12.5 msec. A magnetic tape system stored an additional 100,000 words. The principal input and output device was the Flexowriter, a typewriter-like device that could store or read keystrokes on strips of paper tape. It operated at about the speeds of an ordinary electric typewriter, from which it was derived. In keeping with its aerospace roots, Computer Research Corporation also offered a converter to enter graphical or other analog data into the machine.⁹³ It was also possible to connect an IBM card reader or punch to the computer. The computer’s operating speed was estimated at about eleven multiplications per second.⁹⁴ The 102A was a well-balanced computer and sold in modest numbers. In 1954 the National Cash Register Company purchased CRC, and the 102 formed the basis of NCR’s entry into the computer business.⁹⁵

1950 年至 1954 年间，Computer Research 的经历仅略有不同。通常，一家小型工程公司会围绕磁鼓存储器设计计算机。输入/输出将由标准 Flexowriter 或从 IBM 租用的打孔卡机处理。然后，该公司将在无线电工程师协会/计算机协会联合计算机会议上宣布新机器。然后，他们会从空军或其他军事机构获得一些订单或开发资金。即使这会带来一些民用订单和适度的生产运行，该公司仍然缺乏资源来应对更大批量或先进的后续设计。最后，一家大型老牌公司会收购这家陷入困境的公司，然后该公司将成为大公司进军计算领域的入口。

Computer Research’s experience was repeated with only minor variations between 1950 and 1954. Typically, a small engineering company would design a computer around a drum memory. I/O would be handled by a standard Flexowriter, or by punched card machines leased from IBM. The company would then announce the new machine at one of the Joint Computer Conferences of the Institute of Radio Engineers/Association for Computing Machinery. They would then get a few orders or development funds from the Air Force or another military agency. Even though that would lead to some civilian orders and modest productions runs, the company would still lack the resources to gear up for greater volume or advanced follow-on designs. Finally, a large, established company would buy the struggling firm, which would then serve as the larger company’s entree into computing.

这些计算机中有许多性能良好，物有所值，但无法避免磁鼓内存固有的速度缓慢。它们的输入/输出设备也带来了困境。Flexowriter 价格便宜，但速度慢。连接打孔卡设备意味着很大一部分利润将直接流向 IBM，而不是陷入困境的新计算机公司。

Many of these computers performed well and represented a good value for the money, but there was no getting around the inherent slowness of the drum memory. Their input/output facilities also presented a dilemma. The Flexowriter was cheap, but slow. Attaching punched card equipment meant that a significant portion of the profits would go directly to IBM, and not to the struggling new computer company.

如上所述，National Cash Register 收购了 CRC。由原 UNIVAC 团队的 Samuel Lubkin 创立的电子计算机公司与以打字机闻名的 Underwood 公司合并。（Underwood 于 1957 年退出计算机行业。）加利福尼亚州帕萨迪纳的联合工程公司于 1956 年被 Burroughs 收购。鼓式计算机的主要遗产可能是它们作为许多商业机器公司进入计算机行业的载体。

As mentioned, National Cash Register bought CRC. Electronic Computer Corporation, founded by Samuel Lubkin of the original UNIVAC team, merged with Underwood Corporation, known for its typewriters. (Underwood left the computer business in 1957.) Consolidated Engineering of Pasadena, California, was absorbed by Burroughs in 1956. The principal legacy of the drum computers may have been their role as the vehicle by which many of the business machine companies entered the computer business.

表 1.2列出了 1952 年中期宣布或上市的其他几种磁鼓计算机。对于这些系统中的每一个，基本系统的基本成本为 65,000 至 85,000 美元（不包括添加的内存、安装或辅助 I/O 设备）。

Table 1.2 lists several other magnetic drum computers announced or available by mid-1952. For each of these systems, the basic cost was from $65,000 to $85,000 for a basic system exclusive of added memory, installation, or auxiliary I/O equipment.

表 1.2

1952 年中期市售的小型计算机

Table 1.2

Commercially available small computers, ca. mid-1952

电脑	字长	内存容量（字）	速度倍数/秒）	制造商
公元 30-201	12 月 10 日	4000	118	加利福尼亚州帕萨迪纳联合工程公司
圆圈	40 位	1024	20	纽约州纽约市 Hogan 实验室
艾利康 100	30 位	512	20	电子计算机公司纽约布鲁克林
微型计算机	12 月 10 日	4096	73	物理研究实验室加利福尼亚州帕萨迪纳
蒙倫博	12 月 20 日	100	2	新泽西州奥兰治门罗计算机公司

来源：数据来自美国海军、海军数学计算咨询小组、中等价位商用通用电子数字计算机研讨会（华盛顿特区，1952 年 5 月 14 日）。

Source : Data from U.S. Navy, Navy Mathematical Computing Advisory Panel, Symposium on Commercially Available General-Purpose Electronic Digital Computers of Moderate Price (Washington, DC, 14 May 1952).

后期鼓机，1953–1956

Later Drum Machines, 1953–1956

LGP-30

20 世纪 50 年代中期，第二波设计更精良的鼓式计算机问世，销量大幅增长。它们为许多既不需要也没有资源购买或租赁大型电子计算机的客户提供了一种实用而可靠的替代方案。

In the mid-1950s a second wave of better-engineered drum computers appeared, and these sold in much larger quantities. They provided a practical and serious alternative for many customers who had neither the need nor the resources to buy or lease a large electronic computer.

1956 年推出的 Librascope/General Precision LGP-30 代表了存储程序计算机的最低设计，至少在十年后微型计算机出现之前是如此。它是一台二进制机器，字长为 30 位，指令库只有 16 条。它的磁鼓可容纳 4,096 个字，平均访问时间约为 2.3 毫秒。输入和输出通过 Flexowriter 进行。

The Librascope/General Precision LGP-30, delivered in 1956, represented a minimum design for a stored-program computer, at least until the minicomputer appeared ten years later. It was a binary machine, with a 30-bit word length and a repertoire of only sixteen instructions. Its drum held 4,096 words, with an average access time of around 2.3 msec. Input and output was through a Flexowriter.

LGP-30 只有 113 个真空管和 1,350 个二极管（不像 UNIVAC 有 5,400 个真空管和 18,000 个二极管），看上去就像一张超大办公桌。它的基本但完整的系统售价为 30,000 美元，也是有史以来最便宜的早期计算机之一。大约生产和销售了 400 台。96^它不是 20 世纪 60 年代后期彻底改变计算的小型计算机的直接祖先，但许多小型计算机先驱都知道 LGP-30。Librascope 在 1962 年推出了晶体管版本，但很快就放弃了通用领域，转向为航空航天和国防客户提供专门的制导和控制计算机。

The LGP-30 had only 113 vacuum tubes and 1,350 diodes (unlike the UNIVAC’s 5,400 tubes and 18,000 diodes), and looked like an oversized office desk. At $30,000 for a basic but complete system, it was also one of the cheapest early computers ever offered. About 400 were produced and sold.⁹⁶ It was not the direct ancestor of the minicomputer, which revolutionized computing in the late 1960s, but many minicomputer pioneers knew of the LGP-30. Librascope offered a transistorized version in 1962, but soon abandoned the general-purpose field and turned to specialized guidance-and-control computers for aerospace and defense customers.

本迪克斯 G-15

Bendix G-15

G-15 由哈里·赫斯基设计，本迪克斯制造，可能是美国制造的唯一一台受到艾伦·图灵而非约翰·冯·诺依曼设计理念显著影响的计算机。两者都提倡存储程序原则，并规定根据先前计算的结果对指令进行条件分支。然而，对于冯·诺依曼来说，基本概念是稳定的线性指令流，偶尔会根据条件测试进行分支。另一方面，图灵认为指令没有基本的线性顺序；对他来说，每个命令都代表某种控制转移^。97

The G-15, designed by Harry Huskey and built by Bendix, was perhaps the only computer built in the United States to have been significantly influenced by the design ideas of Alan Turing rather than John von Neumann. Both advocated the stored-program principle, with a provision for conditional branching of instructions based on previously calculated results. For von Neumann, however, the fundamental concept was of a steady linear stream of instructions that occasionally branched based on a conditional test. Turing, on the other hand, felt that there was no fundamental linear order to instructions; for him, every order represented a transfer of control of some sort.⁹⁷

图灵的概念（此处简化了很多）比线性模型更微妙，并且与基于鼓的计算机的性质非常吻合。图灵的模型要求每条指令都带有下一条指令所在的地址，而不是假设下一条指令位于下一个地址位置。在鼓式计算机中，将指令一个接一个地排列是不切实际的，因为这可能要求鼓几乎旋转一整圈，下一条指令才会出现在读取头下。鼓式计算机的程序员通常会开发复杂的“最小延迟编码”方案，将指令分散在鼓表面周围，以确保下一条指令在需要时靠近读取头。（请注意，如果使用访问每条数据所用时间相同的内存，则无需执行上述任何操作。）

Turing’s concept (much simplified here) was more subtle than the linear model, and fit well with the nature of drum-based computers. Turing’s model required that every instruction have with it the address where the next instruction was located, rather than assuming that the next instruction would be found in the very next address location. In a drum computer, it was not practical to have instructions arranged one right after the other, since that might require almost a full revolution of the drum before the next one appeared under the read head. Programmers of drum computers often developed complicated “minimum latency coding” schemes to scatter instructions around the drum surface, to ensure that the next instruction would be close to the read head when it was needed. (Note that none of this was required if a memory that took the same amount of time to access each piece of data was used.)

哈里·赫斯基曾于 1947 年与图灵一起在英国国家物理实验室参与 ACE 项目，1953 年他在底特律韦恩州立大学工作期间设计了 G-15。第一批产品于 1956 年交付，基本价格为 45,000 美元。这款产品被认为很难编程，但对于那些会编程的人来说，它运行速度非常快。本迪克斯销售了四百多台机器，但 G-15 的成功不足以让本迪克斯成为计算机领域的主要参与者。98^控制数据公司后来接管了本迪克斯的计算机业务，本迪克斯继续只提供航空电子设备和国防电子系统。

Harry Huskey, who had worked with Turing in 1947 on the ACE project at the National Physical Laboratory in England, designed what became the G-15 while at Wayne State University in Detroit in 1953. First deliveries were in 1956, at a basic price of $45,000. It was regarded as difficult to program, but for those who could program it, it was very fast. Bendix sold more than four-hundred machines, but the G-15’s success was not sufficient to establish Bendix as a major player in the computer field.⁹⁸ Control Data Corporation later took over Bendix’s computer business, and Bendix continued to supply only avionics and defense electronics systems.

IBM 650

除了国防计算器（又名 IBM 701），IBM 还在开发一款更普通的电子计算机。这台机器起源于 IBM 在其位于纽约州恩迪科特的工厂开发的打孔卡设备扩展提案。IBM 的内部管理层对这个项目犹豫不决，也没有就机器的类型达成一致。一项被称为“木轮”的提案是用于像 604 乘法器这样的插入式编程机器。99^在开发过程中，设计转向使用磁鼓作为主存储器的通用存储程序计算机。（IBM 于 1949 年从工程研究协会收购了磁鼓存储器技术，这是这一转变的关键因素。100 ）这台名为^{650的机器于 1954 年交付，并被证明非常成功，最终安装了大约}^一千台，每月租金约为 3,500 美元。101

Along with the Defense Calculator (a.k.a. IBM 701), IBM was working on a more modest electronic computer. This machine had its origins in proposals for extensions of punched card equipment, which IBM had been developing at its Endicott, New York, plant. IBM’s internal management was hesitant about this project, nor was there agreement as to what kind of machine it would be. One proposal, dubbed “Wooden Wheel,” was for a plug-programmed machine like the 604 Multiplier.⁹⁹ In the course of its development, the design shifted to a general-purpose, stored-program computer that used a magnetic drum for primary memory. (IBM’s acquisition, in 1949, of drum-memory technology from Engineering Research Associates was a key element in this shift.¹⁰⁰) The machine, called the 650, was delivered in 1954 and proved very successful, with eventually around a thousand installations at a rental of around $3,500 a month.¹⁰¹

到发布时，650 必须与许多其他廉价的鼓式机器竞争。它的销量超过了所有机器，部分原因是 IBM 的声誉和庞大的打孔卡用户群，部分原因是 650 被认为比其竞争对手更容易编程和更可靠。IBM 销售人员还迅速指出，650 的鼓式访问时间（2.4 毫秒）比其他鼓式机器（Bendix G-15 除外）更快^。102

By the time of its announcement, the 650 had to compete with many other inexpensive drum machines. It outsold them all, in part because of IBM’s reputation and large customer base of punched card users, and in part because the 650 was perceived as easier to program and more reliable than its competitors. IBM salesmen were also quick to point out that the 650’s drum had a faster access time (2.4 msec) than other drum machines (except the Bendix G-15).¹⁰²

650 被定位为商用机器，并延续了 IBM 为商业和科学客户提供两条不同产品线的政策。具有讽刺意味的是，它对商业客户的影响不如对大学的影响大，而商业客户正是它的目标客户。Thomas Watson Jr. 指示，如果大学同意开设商业数据处理或科学计算课程，IBM 允许大学以高达 60% 的折扣购买 650。许多大学接受了这一提议，使得 650 成为 20 世纪 50 年代后期新兴“计算机科学”部门可用的第一台机器^。103

The 650 was positioned as a business machine and continued IBM’s policy of offering two distinct lines of products for business and scientific customers. Ironically, it had less impact among business customers, for whom it was intended, than it had at universities. Thomas Watson Jr. directed that IBM allow universities to acquire a 650 at up to a 60 percent discount, if the university agreed to offer courses in business data processing or scientific computing. Many universities took up this offer, making the 650 the first machine available to nascent “computer science” departments in the late 1950s.¹⁰³

概括

Summary

在我们讨论的这段时间里，这些机器（无论由谁制造）很少售出。其中大部分（我猜至少 80%）都是由购买的客户而不是销售人员购买的，尽管销售人员可能会获得佣金^。104

Very few of these machines of anybody’s manufacture were sold during the period we are talking about. Most of them, and I would guess 80 percent at least, were bought by the customer who made the buy, not the salesman who made the sale, although the salesman might get the commission.¹⁰⁴

—兰斯洛特·阿姆斯特朗

— Lancelot Armstrong

“第一代”始于商业计算机的推出，当时生产和销售的数量很少。这一阶段始于 1950 年左右，持续了整个十年。这个时代的计算机将程序存储在内部，并使用真空管作为交换技术，但除此之外，它们几乎没有其他共同点。处理器的内部设计差异很大。是否用二进制编码每个十进制数字还是在内部完全以二进制系统运行仍是一个悬而未决的问题。最大的差异在于用于内存的设备：延迟线、威廉姆斯管或鼓。由于所有这些技术在某种程度上都不令人满意，因此人们制造了各种倾向于一种设计方法而不是另一种设计的机器。

The “first generation” began with the introduction of commercial computers manufactured and sold in modest quantities. This phase began around 1950 and lasted through the decade. Computers of this era stored their programs internally and used vacuum tubes as their switching technology, but beyond that there were few other things they had in common. The internal design of the processors varied widely. Whether to code each decimal digit in binary or operate entirely in the binary system internally remained an unsettled question. The greatest variation was found in the devices used for memory: delay line, Williams tube, or drum. Because in one way or another all these techniques were unsatisfactory, a variety of machines that favored one design approach over another were built.

高等研究院的报告由亚瑟·伯克斯 (Arthur Burks)、赫尔曼·戈德斯坦 (Herman Goldstine) 和约翰·冯·诺依曼 (John von Neumann) 撰写，强调了纯二进制设计的优势，使用 RCA 公司设计的名为 Selectron 的存储设备，并行存储器可以一次读取和写入一个字的所有位。然而，当 RCA 公司能够生产足够数量的 Selectron 时，核心存储器已经推出，Selectron 不再那么有吸引力。只有兰德公司制造的 Johnniac 使用了它。大多数其他并行字计算机都使用 Williams 管。105^实际上，这些管子存在可靠性问题^。106

The Institute for Advanced Study’s reports, written by Arthur Burks, Herman Goldstine, and John von Neumann, emphasized the advantages of a pure binary design, with a parallel memory that could read and write all the bits of a word at once, using a storage device designed at RCA called the Selectron. By the time RCA was able to produce sufficient quantities of Selectrons, however, core memory was being introduced, and the Selectron no longer looked so attractive. Only the Johnniac, built at the RAND Corporation, used it. Most of the other parallel-word computers used Williams Tubes.¹⁰⁵ In practice, these tubes were plagued by reliability problems.¹⁰⁶

结果，大多数第一代计算机都使用了每次串行访问一位的存储设备。最快的计算机使用汞延迟线，但最流行的设备是旋转磁鼓。磁鼓本质上是一种机电设备，本质上很慢，但其可靠性和低成本使其成为小型机器的首选技术。

The result was that memory devices that accessed bits one at a time, serially, were used in most first-generation computers. The fastest computers used mercury delay lines, but the most popular device was the rotating magnetic drum. A drum is fundamentally an electromechanical device and by nature slow, but its reliability and low cost made it the technology of choice for small-scale machines.

20 世纪 50 年代初，商业计算起步不顺。Eckert 和 Mauchly 对其潜力有着清晰的认识，为了生存，他们不得不将业务出售给 Remington Rand，Engineering Research Associates 也是如此。然而，Remington Rand 并不完全了解自己购买的是什么。IBM 知道计算机是一件值得参与的事情，但它不确定这些昂贵而复杂的机器如何融入其成功的制表设备系列。客户主动寻找供应商，也许是在参加了 1946 年的 Moore School 课程或参观了正在制造冯·诺依曼型机器的大学之后。这些来自不同背景的客户都迫切需要计算机，尽管 UNIVAC 或 IBM 销售人员不愿意销售计算机。

Commercial computing got off to a shaky start in the early 1950s. Eckert and Mauchly, who had a clear vision of its potential, had to sell their business to Remington Rand to survive, as did Engineering Research Associates. Remington Rand, however, did not fully understand what it had bought. IBM knew that computers were something to be involved with, but it was not sure how these expensive and complex machines might fit into its successful line of tabulating equipment. Customers took the initiative and sought out suppliers, perhaps after attending the Moore School session in 1946 or visiting a university where a von Neumann type machine was being built. These customers, from a variety of backgrounds, clamored for computers, in spite of a reluctance among UNIVAC or IBM salesmen to sell them.

UNIVAC 和 IBM 701 开创了商业存储程序计算时代。它们各有缺点，但总体上满足了订购它们的客户的期望。UNIVAC 的内存可靠但速度慢；701 的内存不太可靠但速度更快。每台机器都运行良好，足以证明大型计算机的可行性。磁鼓技术以较低的每比特成本提供存储，但其速度要慢两个数量级，更接近卡片编程计算器（每分钟能够读取 125 张指令卡）的速度，后者自 20 世纪 40 年代末以来一直由 IBM 提供。考虑到速度方面的劣势，无论价格如何，基于磁鼓的计算机都永远无法与其他计算机竞争。存储程序电子计算机架构在 20 世纪 40 年代承诺的诸多好处需要大容量、高速的内存来匹配电子处理。随着铁氧体磁芯的出现（以及大批量制造它们的技术），第一代计算机的内存问题得到了有效解决。

The UNIVAC and the IBM 701 inaugurated the era of commercial stored-program computing. Each had its drawbacks, but overall they met the expectations of the customers who ordered them. The UNIVAC’s memory was reliable but slow; the 701’s was less reliable but faster. Each machine worked well enough to establish the viability of large computers. Drum technology was providing storage at a lower cost per bit, but its speed was two orders of magnitude slower, closer to the speeds of the Card-Programmed Calculator (which was capable of reading 125 instruction cards per minute), which had been available since the late 1940s from IBM. Given the speed penalty, drum-based computers would never be able to compete with the others, regardless of price. The many benefits promised in the 1940s by the stored-program electronic computer architecture required high-capacity, high-speed memory to match electronic processing. With the advent of ferrite cores—and techniques for manufacturing them in large quantities—the memory problem that characterized the first generation was effectively solved.

表 1.3

早期商用计算机的部分特征

Table 1.3

Selected characteristics of early commercial computers

电脑	字长	内存容量（字）	访问时间（微秒）	每秒乘法次数
CRC-102	12 月 9 日	1024	12,500	65
时代1103	36 位	1024	10	2500–8000
G-15	29 位	2160	平均 1,700	600
LGP-30	30 位	4096	平均 8,500	60
IBM 650	12 月 10 日	1000–2000	平均 2,400	50–450
IBM 701	36 位	2048	四十八	2000
尤尼维克	12 月 11 日	1000	最多 400。	465

来源：数据来自马丁·维克（Martin Weik），《电子数字计算系统调查》，弹道研究实验室报告#971（马里兰州阿伯丁试验场，1955 年 12 月）。

Source: Data from Martin Weik, “A Survey of Electronic Digital Computing Systems,” Ballistic Research Laboratories Report #971 (Aberdeen Proving Ground, Maryland, December 1955).

表 1.3列出了这个时代主要计算机的内存和处理器特性。

Table 1.3 lists memory and processor characteristics of the major computers of this era.

计算技术成熟，1956-1964 年

Computing Comes of Age, 1956–1964

计算机技术渗透到美国每个人的日常生活中。航空旅客的机票、座位分配和账单都由复杂的在线预订系统处理。汽车司机由保险公司投保，保险公司将每位司机的详细和准确的保单记录保存在一个大型数据库中。支票由计算机处理，计算机可以读取支票底部用特殊墨水写的数字。每年四月，公民提交复杂的纳税申报表，美国国税局用计算机处理、归档和跟踪这些纳税申报表。

Computer technology pervades the daily life of everyone in the United States. An airline traveler’s tickets, seat assignment, and billing are handled by a sophisticated on-line reservation system. Those who drive a car are insured by a company that keeps a detailed and exacting record of each driver’s policy in a large database. Checks are processed by computers that read the numerals written in special ink at the bottom. Each April, citizens file complicated tax returns, which the Internal Revenue Service processes, files, and keeps track of with computers.

很难想象一个没有计算机帮助完成这些活动的世界，但直到 20 世纪 50 年代末，计算机才被计算机化。这为 20 年后计算机的进一步渗透奠定了基础，包括自动取款机、超市和零售收银台扫描的条形码产品，以及信用卡公司和邮购公司维护的大量财务和个人数据库。

It is hard to imagine a world in which computers do not assist with these activities, yet they were not computerized until the late 1950s. This set the stage for further penetration of computing two decades later, in the form of automatic teller machines, bar-coded products scanned at supermarket and retail check-out stations, and massive financial and personal databases maintained by credit-card companies and mail-order houses.

1955 年之前，人类使用打字机、复写纸和大量文件柜来完成所有这些活动。1打孔卡设备协助完成部分工作。最受欢迎的算术辅助工具是芝加哥 Felt and Tarrant 公司制造的 Comptometer（^图2.1）。2^这台机器是按键驱动的：按下按键即可立即执行加法，无需拉动其他杠杆或按下其他按钮。使用它需要经过强化训练，但在熟练操作员的手中，Comptometer 可以每隔几秒钟执行一次加法。但是，它既不能进行乘法，也不能打印计算结果。

Before 1955, human beings performed all these activities using typewriters, carbon paper, and lots of filing cabinets.¹ Punched-card equipment assisted with some of the work. The preferred aid to arithmetic was the Comptometer, manufactured by Felt and Tarrant of Chicago (figure 2.1).² This machine was key-driven: pressing the keys immediately performed the addition, with no other levers to pull or buttons to press. Its use required intensive training, but in the hands of a skilled operator, a Comptometer could perform an addition every few seconds. It could neither multiply nor print the results of a calculation, however.

这些应用程序的共同点是它们需要轻松快速地存储和检索大量数据。还需要各种检索方法，以便以后可以以不同的方式使用数据。计算主要包括加法、减法，以及不太常见的乘法。数量通常高达一百万，并且需要精确到小数点后两位（以美元和美分表示）。虽然与打孔卡设备处理的工作类似，但这项活动还要求从大文件中快速检索单个记录，这是打孔卡机器无法轻松完成的。“数据处理”的定义不断发展以适应这一变化。

What these applications had in common was their need to store and retrieve large amounts of data easily and quickly. Required also were a variety of retrieval methods, so that the data could be used later on in different ways. Calculations consisted mainly of additions, subtractions, and less frequently, multiplications. Quantities typically ranged up to a million and required a precision of two decimal places, for dollars and cents. Though similar to the work that punched card installations handled, this activity had the additional requirement of rapid retrieval of individual records from large files, something punched card machines could not easily do. The definition of “data processing” evolved to accommodate this change.

图 2.1

Comptometer。（来源：史密森学会。）

Figure 2.1

Comptometer. (Source : Smithsonian Institution.)

20 世纪 50 年代初期的计算机并不适合这项工作。当时大量出现的廉价鼓式计算机缺乏内存容量、速度，尤其是高容量输入和输出设备。大型计算机显示出更大的潜力，但即使是从一开始就为数据处理应用而设计的 UNIVAC，在首次推出时也配备了速度很慢的打印机。

The computers of the early 1950s were ill suited for this work. The inexpensive drum-based machines that proliferated early in the decade lacked the memory capacity, speed, and above all, high-capacity input and output facilities. The larger machines showed more potential, but even the UNIVAC, designed for data processing applications from the start, had a slow printer when first introduced.

到 20 世纪 50 年代末，数字电子计算机开始做这种工作。整个 20 世纪 50 年代，计算机设计师将为解决科学问题而开发的机器的架构调整到需要更多存储空间和更大量输入和输出的应用程序。这些都是根本性的变化，但计算机在发展过程中适应了这些变化，同时又不放弃其基本的存储程序架构。

By the end of the 1950s, digital electronic computers had begun doing that kind of work. Through the 1950s, computer designers adapted the architecture of a machine developed for scientific problems to applications that required more storage and more voluminous input and output. These were fundamental changes, but computers evolved to accommodate them without abandoning their basic stored-program architecture.

核心内存

Core Memory

计算机的这种转变部分源于电路技术的进步。到 1959 年，晶体管已经变得足够可靠和便宜，可以作为处理器的基本电路元件。结果，可靠性提高了，维护成本降低了，运营成本也降低了。然而，在此之前，出现了一项更为激进的创新——开发出由磁芯制成的可靠、高容量内存单元。这两项创新能够将性能提升到许多商业应用变得具有成本效益的程度。

Part of this transformation of computers came from advances in circuit technology. By 1959 the transistor had become reliable and cheap enough to serve as the basic circuit element for processors. The result was increased reliability, lower maintenance, and lower operating costs. Before that, however, an even more radical innovation occurred—the development of reliable, high capacity memory units built out of magnetic cores. These two innovations were able to boost performance to a point where many commercial applications became cost-effective.

磁芯存储器是指一些小的、甜甜圈状的材料，几根细线穿过这些材料来存储信息（图 2.2）。穿过磁芯的细线可以使磁芯在任一方向上磁化；另一根穿过的细线可以感应到这个方向，这个方向被定义为二进制的零或一。该技术利用了某些磁性材料的一种称为磁滞的特性。只有当电流超过某个阈值时，流过这种磁芯中心的电流才会磁化。3^同样，如果电流足够强，流过另一个方向的电流也会使这种磁芯消磁。磁芯存储器将由具有这种特性的材料制成的磁芯排列在一个平面上，细线垂直和水平地穿过每个磁芯的孔。只有当垂直和水平细线上都有电流，且两者都沿同一方向流动时，磁芯才会被磁化；否则，不会产生任何效果。

Core memory refers to small, doughnut-shaped pieces of material through which several fine wires are threaded to store information (figure 2.2). The wires passing through the core can magnetize it in either direction; this direction, which another wire passing through can sense, is defined as a binary zero or one. The technology exploits the property, known as hysteresis, of certain magnetic materials. A current passing through the center of such a core will magnetize it, but only if it is above a certain threshold.³ Likewise, a current passing in the other direction will demagnetize such a core if the current is strong enough. A core memory unit arranges cores made of materials having this property in a plane, with wires running vertically and horizontally through the hole in each core. Only when there are currents in both the vertical and the horizontal wires, and both are running in the same direction, will a core be magnetized; otherwise, there is no effect.

与第一代计算机中使用的存储器相比，磁芯存储器具有许多优势。磁芯可以做得很小。该存储器是“非易失性的”：它无需提供电力（如威廉姆斯管和汞延迟线）或机械能（如旋转鼓）即可保存信息。

A core memory has many advantages over the memories used in the first-generation computers. The cores can be made small. The memory is “nonvolatile”: it holds information without having to supply electrical power (as with Williams tubes and mercury delay lines) or mechanical power (as with a rotating drum).

最重要的是，核心提供随机存取存储器，现在称为 RAM：访问核心平面的任何位都与访问其他位一样快。（这个术语具有误导性：它实际上不是“随机”时间，但由于该术语很常用，因此将在此处保留。）这克服了延迟线和延迟鼓的主要缺点，即等待数据到达会导致延迟，从而降低计算机速度。

Above all, core provides random access memory, now known as RAM: access to any bit of a core plane is as rapid as to any other. (The term is misleading: it is not really a “random” time, but since the term is in common use it will be retained here.) This overcomes a major drawback to delay lines and drums, where waiting for data to come around can introduce a delay that slows a computer down.

第二次世界大战期间，德国海军开发了一种具有磁滞特性的磁性材料，并将其用于火控系统的电路中。战后，样品被带到了美国，引起了对数字存储感兴趣的人的注意。IBM、伊利诺伊大学、哈佛大学、麻省理工学院等机构的研究人员研究了它对计算机的适用性。4 哈佛大学^霍华德·艾肯的学生王安发明了一种磁芯存储器，用于 1952 年完成的哈佛 Mark IV。1953 年夏天，ENIAC 和 Whirlwind 都安装了磁芯存储器。ENIAC 的存储器由 Burroughs 公司设计，使用二维磁芯阵列；Whirlwind 的存储器由 Jay Forrester 设计，使用三维阵列，可提供更快的切换速度、更大的存储密度和更简单的电子设备。⁵ Forrester 设计的一个关键优势是采用了由 Ken Olsen 开发的电路，该电路减少了操作阵列所需的电流量。

During World War II, the German Navy developed a magnetic material with the property of hysteresis, and they used it in the circuits of a fire-control system. After the war, samples were brought to the United States, where it caught the attention of people interested in digital storage. Researchers at IBM, the University of Illinois, Harvard, MIT, and elsewhere investigated its suitability for computers.⁴ An Wang, a student of Howard Aiken at Harvard, invented a core memory that was used in the Harvard Mark IV, completed in 1952. Magnetic core memories were installed on both the ENIAC and the Whirlwind in the summer of 1953. The ENIAC’s memory, designed by the Burroughs Corporation, used a two-dimensional array of cores; the Whirlwind’s memory, designed by Jay Forrester, used a three-dimensional array that offered faster switching speeds, greater storage density, and simpler electronics.⁵ One key advantage of Forrester’s design was a circuit, developed by Ken Olsen, that reduced the amount of current needed to operate the array.

图 2.2

磁芯存储器。（来源：Jan A. Rajchman，《万比特磁芯矩阵存储器》， IRE 会议录（1953 年 10 月）：1408。）©1953 IRE，现称为 IEEE。

Figure 2.2

Magnetic core memory. (Source : From Jan A. Rajchman, “A Myriabit Magnetic-Core Matrix Memory,” IRE Proceedings (October 1953): 1408.) ©1953 IRE, now known as IEEE.

核心内存使 Whirlwind 几乎成为一台新机器，性能大大提升，商业系统也开始采用它。如上一章所述，第一次商业交付是在 1954 年末，当时雷明顿兰德的 ERA 部门向国家航空咨询委员会交付了一台 1103A 计算机。ERA 还向国家安全局交付了核心内存，作为机密项目的一部分。在 IBM，由 Eric Bloch 领导的团队开发了一种内存单元，用作 702 计算机的静电内存与其基于卡的输入和输出单元之间的缓冲器。向商业客户的交付始于 1955 年 2 月。IBM 继续为 702 使用静电管，但在之后制造的机器中转向核心^。6

The core memory made the Whirlwind almost a new machine, so much better was its performance, and commercial systems began appearing with it. As mentioned in the previous chapter, the first commercial delivery was around late 1954, when the ERA division of Remington Rand delivered an 1103A computer to the National Advisory Committee for Aeronautics. ERA had also delivered core memories to the National Security Agency as part of a classified project. At IBM, a team led by Eric Bloch developed a memory unit that served as a buffer between the electrostatic memory of the 702 computer and its card-based input and output units. Deliveries to commercial customers began in February 1955. IBM continued using electrostatic tubes for the 702 but moved to core for machines built after it.⁶

与美国空军签订的旋风生产版合同至关重要，因为它为工程师提供了将核心系统应用于商业系统所需的经验。空军的 SAGE（半自动地面环境）是一个集计算机、雷达、飞机、电话线、无线电链路和船舶于一体的系统，旨在探测、识别和协助拦截试图侵入美国上空的敌机。它的核心是一台计算机，它将协调从遥远来源收集的信息，对其进行处理，并以文本和图形形式呈现。总而言之，这是一个雄心勃勃的设计；空军希望全天候运行多台这台计算机，这让它更加雄心勃勃。7^该系统的主要要求是高可靠性，这就排除了水银延迟线或静电存储器。

A contract with the U.S. Air Force to build production versions of the Whirlwind was a crucial event because it gave engineers the experience needed for core to become viable in commercial systems. The Air Force’s SAGE (Semi-Automatic Ground Environment), a system that combined computers, radar, aircraft, telephone lines, radio links, and ships, was intended to detect, identify, and assist the interception of enemy aircraft attempting to penetrate the skies over the United States. At its center was a computer that would coordinate the information gathered from far-flung sources, process it, and present it in a combination of textual and graphical form. All in all, it was an ambitious design; the Air Force’s desire to have multiple copies of this computer in operation round the clock made it even more so.⁷ A primary requirement for the system was high reliability, which ruled out mercury delay lines or electrostatic memory.

SAGE 计算机的设计与 Whirlwind 有很多共同之处；一些早期文献将其描述为“Whirlwind II”。这在其核心内存中尤为明显，其设计容量为 8,192 个字，长度为 32 位。1952 年，林肯实验室的 SAGE 开发团队向三家公司询问了当时正在设计的计算机的生产模型的可能性。该团队参观了 IBM、Raytheon 和 Remington Rand 的两个部门的设施。在对工厂进行全面评估后，该团队选择了 IBM。8 ^IBM于 1955 年交付了原型，并于次年完成了第一台生产模型计算机。IBM 最终为 SAGE 交付了大约 30 个计算机系统。为了提高可靠性，每个系统都由两台串联运行的相同计算机组成，如果主计算机发生故障，则有一个开关可立即将控制权转移到备用计算机。尽管这些计算机使用了真空管（每对 55,000 个），但双工系统的可靠性超过了多年后制造的大多数固态计算机。最后一台原始的 SAGE 计算机位于安大略省北湾的一个站点，于 1983 年关闭^。9

The design of SAGE’s computer had much in common with Whirlwind; some early literature described it as “Whirlwind II.” That was especially evident in its core memory, designed to have a capacity of 8,192 words of 32 bits in length. In 1952 the SAGE development team at Lincoln Laboratory asked three companies about the possibility of building production models of the computer then being designed. The team visited the facilities of IBM, Raytheon, and two divisions of Remington Rand. Based on a thorough evaluation of the plants, the team selected IBM.⁸ IBM delivered a prototype in 1955, and completed the first production model computer the following year. IBM eventually delivered around thirty computer systems for SAGE. For reliability, each system consisted of two identical computers running in tandem, with a switch to transfer control immediately to the backup if the primary computer failed. Although the computers used vacuum tubes (55,000 per pair), the reliability of the duplexed system exceeded that of most solid-state computers built years later. The last original SAGE computer, operating at a site in North Bay, Ontario, was shut down in 1983.⁹

图 2.3

(顶部) 在安装到 Whirlwind 之前为内存测试计算机开发的核心内存单元。（来源：Mitre Corporation 档案。）（底部）IBM 704 计算机的核心内存单元。（来源：明尼苏达大学查尔斯·巴贝奇研究所。）

Figure 2.3

(top) Core memory unit developed for the Memory Test Computer, prior to installation on the Whirlwind. (Source : Mitre Corporation Archives.) (bottom) Core memory unit for the IBM 704 computer. (Source : Charles Babbage Institute, University of Minnesota.)

最初，IBM 与其他公司（主要是 General Ceramics）签订了合同，由后者提供磁芯。此外，IBM 还开始在内部研究磁芯生产。除其他事项外，IBM 还与 Colton Manufacturing Company（为制药行业提供制造药丸的机器）合作，使其设备能够压制具有统一属性的磁芯。（IBM 和其他计算机公司还使用 General Mills 生产的机器改装而成的机器，用于将食品放入消费包装中，以及 United Shoe Machinery 生产的机器改装而成的机器，用于将电子元件插入电路板。）¹⁰随着 SAGE 项目的开展，IBM 开始越来越多地依赖自己的专业知识。SAGE 需要数十万个优质磁芯。鉴于最初提供给 IBM 的磁芯产量较低，似乎必须制造和测试数百万个。11 ^IBM自己的研究工作进展顺利，到 1954 年，产量高达 95%（图 2.3）。

Initially IBM had contracted with other companies, primarily General Ceramics, to deliver cores. It had also begun a research effort on core production in-house. Among other things, it worked with the Colton Manufacturing Company, which provided machines to the pharmaceutical industry for making pills, to adapt their equipment to press cores of uniform properties. (IBM and other computer companies also used machines modified from those made by General Mills for putting food into consumer packages, and by United Shoe Machinery for making shoes, to insert electronic components onto circuit boards.)¹⁰ As the SAGE project got underway, IBM began to rely more and more on its own expertise. SAGE would require hundreds of thousands of good cores. Given the low yields of cores supplied to IBM at first, it seemed that millions would have to be made and tested.¹¹ IBM’s own research efforts fared much better, producing yields of up to 95 percent by 1954 (figure 2.3).

SAGE 合同在 20 世纪 50 年代为 IBM 带来了 5 亿美元的收入。它在使 IBM 进入核心生产业务方面所起的作用可能也同样巨大。12到¹⁹⁵⁶年，IBM 在大型计算机的安装数量上已经超过了 UNIVAC。IBM 已经在小型计算机的销售中占据主导地位，并且将继续主宰整个计算机行业。13它是如何做到^这一点的一直是许多报道的主题。大多数人将功劳归于 IBM 的销售团队，同时也指出雷明顿兰德的高层管理人员对 UNIVAC 部门的支持力度较小。一些报道认为，尽管 UNIVAC 的机器在技术上占有优势，但 IBM 仍然取得了领先地位。14 IBM^从SAGE 的经验中获得了生产可靠核心存储器的经验，这也很重要。

The SAGE contract generated half a billion dollars in revenue for IBM in the 1950s. Its value in getting IBM into the business of producing cores was probably worth just as much.¹² By 1956, IBM had surpassed UNIVAC in the number of installations of large computers. Already dominant in sales of smaller computers, IBM would continue to dominate the entire computer industry.¹³ How it managed to do that has been the subject of many accounts. Most give generous credit to IBM’s sales force, and note also that Remington Rand’s top management was less forceful in their support of the UNIVAC division. Some accounts believe that IBM took this lead despite the technical superiority of UNIVAC’s machines.¹⁴ Also important was the experience in producing reliable core memories that IBM gained from its experience with SAGE.

霍尼韦尔、通用电气、RCA

Honeywell, GE, RCA

当 IBM 和 UNIVAC 争夺业务时，其他公司也开始开发和销售大型计算机。其中首当其冲的是明尼阿波利斯霍尼韦尔公司，这是一家工业和消费控制（包括家用恒温器）和航空电子设备制造商。1955 年，霍尼韦尔收购了雷神公司的计算机部门，雷神公司是唯一一家响应美国政府在 20 世纪 40 年代末的请求，为其提供大型计算机的公司。雷神公司未能交付其承诺的机器，尽管一台名为 RAYDAC 的计算机于 1952 年作为“飓风”项目的一部分安装在加利福尼亚州 Point Mugu 的美国海军基地。1954 年，雷神公司与霍尼韦尔联合成立了 Datamatic 公司，但次年它放弃了对 Datamatic 的所有权益^。15

As IBM and UNIVAC competed for business, other companies took steps to develop and sell large computers. First among them was the Minneapolis Honeywell Corporation, a maker of industrial and consumer controls (including thermostats for the home) and aerospace electronics equipment. In 1955 Honeywell acquired the computer division of Raytheon, which had been the only established company to respond to the U.S. government’s request in the late 1940s for large computers for its needs. Raytheon was unable to deliver the machines it promised, although one computer, the RAYDAC, was installed in 1952 at a U.S. Navy base at Point Mugu, California, as part of Project “Hurricane.” In 1954 Raytheon established the Datamatic Corporation jointly with Honeywell, but the following year it relinquished all its interest in Datamatic.¹⁵

霍尼韦尔的第一款大型产品是 Datamatic 1000，于 1957 年交付。这款机器可与最大的 UNIVAC 或 IBM 系统相媲美，但已经过时了。除其他外，它使用真空管，而当时晶体管的实用性已逐渐明朗。霍尼韦尔暂时退出市场，专注于设计晶体管机器，几年后成功推出。这一决定为其在 20 世纪 60 年代中期成功重返市场奠定了基础^。16

Honeywell’s first large offering was the Datamatic 1000, delivered in 1957. This machine was comparable to the largest UNIVAC or IBM systems, but it was already obsolete. Among other things, it used vacuum tubes at a time when it was becoming clear that transistors were practical. Honeywell temporarily withdrew from the market and concentrated on designing transistorized machines, which it successfully offered a few years later. That decision laid the grounds for its successful reentry, which began in the mid-1960s.¹⁶

通用电气

1955 年，通用电气是美国最大的电子公司，销售额近 30 亿美元，员工超过 200,000 人。（相比之下，IBM 的销售额为 4.61 亿美元，员工人数为 46,500 人，而雷明顿兰德的销售额为 2.25 亿美元，员工人数为 37,000 人。）¹⁷ 1953 年，该公司向赖特-帕特森空军基地的美国空军交付了 OARAC，空军将其用于专门的机密应用。但是，OARAC 是一种通用电子计算机，如果 GE 的高层管理人员没有决定不进入计算机领域，GE 本可以销售其商业版本。GE 工程师后来回忆说，整个 20 世纪 50 年代，GE 一直对进入这个市场存在偏见。GE 表示，它更愿意专注于它认为更有潜力的其他产品，比如喷气发动机和核能。其他人对这种说法提出质疑。¹⁸一位工程师认为，IBM 是 GE 最大的真空管客户这一事实可能是一个因素：GE 不希望表现出与 IBM 的竞争，尤其是考虑到人们认为 GE 拥有更强大的资源，如果选择这样做，可以压倒 IBM。

In 1955, General Electric was the nation’s leading electronics firm, with sales of almost $3 billion and over 200,000 employees. (Compare IBM’s sales of $461 million and 46,500 employees, or Remington Rand’s $225 million and 37,000 employees that year.)¹⁷ In 1953, the company had delivered the OARAC to the U.S. Air Force at Wright-Patterson Air Force Base, and the Air Force had used it for specialized, classified applications. However, the OARAC was a general-purpose electronic computer and GE could have marketed a commercial version of it if its senior management had not decided against entering the computer field. GE engineers later recalled a consistent bias against entering this market throughout the 1950s. GE said that it preferred to concentrate on other products it felt had greater potential, like jet engines and nuclear power. Others dispute that account.¹⁸ One engineer suggested that the fact that IBM was GE’s largest customer for vacuum tubes might have been a factor: GE did not want to appear to be in competition with IBM, especially given the perception that GE, with its greater resources, could overwhelm IBM if it chose to do so.

然而，通用电气在 20 世纪 50 年代末确实为一个名为 ERMA（电子记录机会计）的系统生产了一台计算机，这是一个与斯坦福研究所和美国银行共同开发的自动支票结算系统。考虑到通用电气首席执行官拉尔夫·科迪纳 (Ralph Cordiner) 禁止通用计算机，ERMA 的创建者将该项目作为一个特殊的一次性项目在内部出售。在亚利桑那州凤凰城外建了一座工厂，由 Homer R. Oldfield 领导的通用电气工程师开始工作。虽然通用电气仍然是真空管的主要供应商，但在其庞大的研究机构中，有了解晶体管优势和问题的人。ERMA 计算机将采用晶体管。交付于 1958 年初。美国银行于 1959 年秋季在通用电气发言人罗纳德·里根主持的仪式上公开发布了^ERMA。19

General Electric did, however, produce a computer in the late 1950s for a system called ERMA (Electronic Recording Machine Accounting), an automatic check-clearing system developed with the Stanford Research Institute and the Bank of America. Mindful of the ban by GE chief Ralph Cordiner against general-purpose computers, ERMA’s creators sold the project internally as a special, one-time project. A plant was built outside Phoenix, Arizona, and GE engineers, led by Homer R. Oldfield, got to work. While still a major supplier of vacuum tubes, GE had among its sprawling research facilities people who understood the advantages—and problems—of transistors. The ERMA computer would be transistorized. Deliveries began in early 1958. The Bank of America publicly unveiled ERMA in the fall of 1959, at a ceremony hosted by GE spokesman Ronald Reagan.¹⁹

ERMA 成功地使银行实现了支票结算这一繁琐过程的自动化，从而避免了在战后经济繁荣时期银行面临的文书工作危机。它的组成部分之一是一组数字和控制字符，这些字符用磁性墨水印在每张支票的底部，机器可以读取。它被称为“MICR”。广告公司采用这种字体作为“计算机语言”的象征，有一段时间，这种字体是自动化时代的象征。然而，很少有人意识到，MICR 只规定了十位小数和一些控制字符的形状，而不是字母表的字母。

ERMA sucessfully allowed banks to automate the tedious process of clearing checks, thus avoiding the crisis of paperwork that threatened banks in the booming postwar economy. Among its components was a set of numeric and control characters printed with magnetic ink at the bottom of each check that a machine could read. It was called “MICR.” Advertising agencies adopted the typography as a symbol of “computerese,” and for a while the type was a symbol of the Age of Automation. Few realize, however, that MICR only specified the shapes of the ten decimal digits and some control characters, not the letters of the alphabet.

ERMA 的成功鼓舞了它的创造者。他们继续冒险，开发了其他计算机，包括 1962 年达特茅斯学院采用的开创性分时系统（将在第 5 章中详细讨论）。与 ERMA 一样，他们将自己的产品描述为专用设备。但他们的把戏只能到此为止。如果没有公司的全力支持，该公司几乎不可能与 IBM 竞争。从某一方面来看，通用电气的管理层是正确的：尽管其产品质量高、技术创新，但计算机部门从未盈利。1970 年，通用电气以略高于 2 亿美元的价格将计算机业务卖给了霍尼韦尔^。20

ERMA’s success emboldened its creators. They continued their risky game by developing other computers, including a system that in 1962 Dartmouth College would adopt for its pioneering time-sharing system (to be discussed at length in chapter 5). As with ERMA, they described their products as special-purpose equipment. But their charade could only go so far. Without full corporate support the company could hardly expect to compete against IBM. In one respect, GE’s management had been correct: the computer division never was profitable, despite the high quality and technical innovation of its products. In 1970 GE sold the business to Honeywell for a little over $200 million.²⁰

美国无线电公司

RCA

RCA 进入商业计算领域的时间与 GE 不相上下。与 GE 一样，RCA 很早就涉足计算机领域——它曾为 20 世纪 50 年代初在普林斯顿制造的计算机开发过存储管。RCA 的规模不如 GE，但 1955 年，其年销售额是 IBM 的两倍多，是雷明顿兰德的四倍。1955 年 11 月，RCA 宣布推出一款用于商业和数据处理应用的大型计算机 BIZMAC。（参与开发的工程师中有阿诺德·斯皮尔伯格，他是一位才华横溢的工程师，在计算机界的名气被他的儿子、好莱坞电影制片人史蒂文所掩盖。阿诺德·斯皮尔伯格后来搬到了凤凰城，并参与了上述 GE 计算机的开发。）

RCA’s entry into commercial computing paralleled GE’s. Like GE, RCA had been involved at an early date in computers—it had developed a storage tube for a computer built at Princeton in the early 1950s. RCA was not as large as GE but in 1955 it had over twice the annual sales of IBM and four times the sales of Remington Rand. In November 1955 RCA announced a large-scale computer intended for business and data processing applications, the BIZMAC. (Among the engineers who worked on it was Arnold Spielberg, a talented engineer whose fame among computer circles would be eclipsed by that of his son, Steven, the Hollywood filmmaker. Arnold Spielberg later moved to Phoenix and worked on the GE computers described above.)

BIZMAC 使用核心内存，这使它成为首批使用核心内存的商用机器之一。真空管用于逻辑和算术电路。BIZMAC 的销量不佳。只有一个完整的系统安装于底特律的美国陆军设施。一些较小的系统安装在其他地方。一个原因可能是开发时间太长。到 1956 年首次安装时，基于更简单的架构并使用晶体管的新设计有望以更低的成本提供同等的性能^。21

The BIZMAC used core memory, which made it one of the first commercial machines to do so. Vacuum tubes were used for logic and arithmetic circuits. The BIZMAC did not sell well. Only one full system was installed, at a U.S. Army facility in Detroit. A few smaller systems were installed elsewhere. One reason might have been a too-long development time. By the time of its first installations in 1956, new designs based on simpler architectures and using transistors promised to offer the equivalent performance at a lower cost.²¹

BIZMAC 的架构与它所竞争的机器不同，这可能是导致其商业失败的另一个因素。与冯·诺依曼的设计不同，BIZMAC 具有专门的处理单元，用于搜索和排序磁带卷上的数据。大多数当代计算机有多达十几个磁带驱动器用于大容量存储，而 BIZMAC 的设计支持数百个驱动器，所有驱动器都连接到其处理器并受机器控制。这意味着每个磁带卷都将永久安装，几乎不需要操作员像其他计算机一样安装和拆卸磁带。22^继电器系统将特定磁带连接到 BIZMAC 的处理器。连接到主处理器的是一个专用处理器，其唯一功能是排序数据。

The BIZMAC’s architecture was different from the machines it was competing with, which may have been another factor that led to its commercial failure. Unlike the von Neumann design, the BIZMAC had specialized processing units for searching and sorting data on reels of tape. Whereas most contemporary computers had up to a dozen tape drives for mass storage, the BIZMAC was designed to support several hundred drives, all connected to its processor and under machine control. That implied that each reel of tape would be permanently mounted, and there would be little or no need for an operator to mount and demount tapes as there was with other computers.²² A system of relays connected a particular tape to the BIZMAC’s processor. Attached to the main processor was a special-purpose processor whose only function was to sort data.

这种设计似乎使 BIZMAC 特别适合商业数据处理应用，但到 1956 年，技术的进步已经消除了这种架构带来的所有优势。其他制造商已经提供性能大大提高的磁带驱动器，这些驱动器与核心内存和处理速度的进步相结合，使得仅使用少量高速磁带驱动器并雇用人工从磁带库安装和卸载磁带的成本更低。

This design would seem to make the BIZMAC especially suited for business data processing applications, but by 1956 advances in technology had eliminated any advantage gained by this architecture. Other manufacturers were already offering tape drives with much-improved performance, and those drives, combined with advances in core memory and processing speeds, made it cheaper to have only a few high-speed tape drives with human beings employed to mount and demount tapes from a library.

BIZMAC 的失败阻碍了 RCA 进军商业计算领域，但并未终结这一事业。在短暂的停滞之后，该公司推出了一系列设计更为传统的晶体管计算机（301 和 501）；这些产品取得了一定的成功。

The BIZMAC’s failure set back RCA’s entry into commercial computing, but it did not end it. After a brief hiatus, the company responded with a line of transistorized computers of more conventional design (the 301 and 501); these were moderately successful.

BIZMAC 并不是唯一一款探索替代架构的大型计算机。1956 年，UNIVAC（现在是 Sperry Rand 的一个部门，由 Sperry Corporation 和 Remington Rand 合并而成）推出了 UNIVAC File Computer，希望它能成为原 UNIVAC 的低成本继任者。顾名思义，这台机器用于数据处理。它使用磁鼓作为主存储器，这降低了成本，但也损害了性能。它由存储指令和接线板面板组合编程。其设计者认为灵活的输入/输出 (I/O) 设施对商业客户至关重要。23^与BIZMAC 一样，UNIVAC File 无需通过中央处理器即可处理数据。UNIVAC File 与 BIZMAC 一样，在商业上失败了，原因可能相同。这种设计用于数据存储、检索、排序和搜索的机器的概念在整个计算历史中不时出现。但几乎在所有情况下，一台好的通用计算机都会使专用机器退出市场。表 2.1列出了 1955 年美国主要计算机供应商及其收入。

The BIZMAC was not the only large computer to explore an alternate architecture. In 1956 UNIVAC (now a division of Sperry Rand, formed by a merger of the Sperry Corporation and Remington Rand) introduced the UNIVAC File Computer, which it hoped would be a low-cost successor to the original UNIVAC. As the name implied, the machine was intended for data handling. It used a magnetic drum as its main memory, which lowered costs but also compromised performance. It was programmed by a combination of stored instructions and plugboard panels. Its designers felt that flexible input/output (I/O) facilities were critical for commercial customers.²³ As with the BIZMAC, the UNIVAC File could manipulate data without having to go through its central processor. The UNIVAC File, like the BIZMAC, was a commercial failure, probably for the same reasons. This notion of designing a machine for data storage, retrieval, sorting, and searching reappears from time to time throughout the history of computing. But in nearly every case, a good general-purpose computer has driven specialized machines from the market. Table 2.1 lists the major U.S. computer suppliers and their revenues for 1955.

表 2.1

1955 年部分计算机和电子公司的收入

Table 2.1

Revenues of selected computer and electronic companies, 1955

公司	年销售额	净利	雇员
通用电气	29.6 亿美元	2.13亿美元	210,000
西部电气*	15亿美元	5500 万美元	98,000
美国无线电公司	9.4 亿美元	4000 万美元	70,500
IBM	4.61亿美元	4650 万美元	46,500
无碳复写纸	2.59亿美元	1270 万美元	37,000
霍尼韦尔	2.29亿美元	1530 万美元	25,000
雷明顿·兰德**	2.25亿美元	1220 万美元	37,000
雷神公司	1.77亿美元	350 万美元	18,700
巴勒斯	1.69亿美元	780 万美元	20,000

资料来源：数据来自《财富》（1955 年 7 月）。

* 西部电气是 AT&T 的制造部门，由 AT&T 拥有和控制。AT&T 1955 年的总收入超过了 GE、RCA 和 IBM 的总收入。

** 1955 年，雷明顿兰德与斯佩里公司合并，斯佩里公司的销售额为 4.41 亿美元，主要与国防相关。

Source : Data from Fortune (July 1955).

* Western Electric was the manufacturing arm of AT&T, which owned and controlled it. AT&T’s total revenues for 1955 were greater than GE’s, RCA’s, and IBM’s combined.

** In 1955 Remington Rand merged with the Sperry Corporation, a company with $441 million in sales, mostly defense-related.

计算机体系结构入门

A Primer on Computer Architecture

到 1960 年底，美国安装了大约 6,000 台通用电子计算机。24^它们几乎都是 20 世纪 40 年代 EDVAC 和 IAS 计算机项目的后代，存储程序的概念首次出现在这两个项目中。人们经常听说，自冯·诺依曼以来，计算机设计并没有发生任何真正的变化。25^这只是在有限的意义上是正确的——计算机仍然在内部存储程序，并在电路中将存储与算术功能分开。在大多数其他方面，都取得了重大进展。到 1960 年，随着不同供应商试图在市场上建立自己的产品，其中一些创新成为了卖点。最重要的架构特征总结如下^。26

By the end of 1960 there were about 6,000 general-purpose electronic computers installed in the United States.²⁴ Nearly all of them were descendents of the EDVAC and IAS computer projects of the 1940s, where the concept of the stored program first appeared. One often hears that nothing has really changed in computer design since von Neumann.²⁵ That is true only in a restricted sense—computers still store their programs internally and separate storage from arithmetic funtions in their circuits. In most other ways there have been significant advances. By 1960 some of these innovations became selling points as different vendors sought to establish their products in the marketplace. The most important architectural features are summarized here.²⁶

字长引入可靠的核心存储器后，可以按位组提取数据，而不必像延迟线那样一次提取一位。对于进行科学计算的计算机来说，让这组数据与典型计算所需的位数相对应似乎是很自然的事情，例如 7 到 12 位十进制数字。这意味着块大小或字长为 30 到 50 位。较长的字长更适合科学计算，但这会增加设计的复杂性和成本。到 1960 年，其他因素使得字长的决策变得更加困难。

Word Length The introduction of reliable core memory made it practical to fetch data in sets of bits, rather than one bit at a time as required by a delay line. For a computer doing scientific work, it seemed natural to have this set correspond to the number of digits required for a typical computation—say, from 7 to 12 decimal digits. That meant a block size, or word length, of from 30 to 50 bits. Longer word lengths were preferred for scientific calculations but increased the complexity and cost of the design. By 1960, additional factors made the word-length decision even more difficult.

用于商业用途的计算机不需要处理多位数字。20 世纪 50 年代处理的货币很少超过一百万美元，小数点右边两位数就足够了。因此，如果有办法告诉处理器何时到达字尾，那么面向商业的计算机可以使用较短的字长或可变的字长。IBM 702、IBM 1401、RCA 301 和 RCA 501 具有可变的字长，字尾通过各种方式设置。1401 使用附加在每个编码字符后的额外位来指示它是否是字的最后一个字符；702 使用一个特殊字符来表示已到达字尾。27 尽管^在20 世纪 50 年代很流行，但具有可变字长的计算机在接下来的几十年里逐渐失宠，现在已不常见。

Computers intended for commercial use did not need to handle numbers with many digits. Money handled in the 1950s seldom exceeded a million dollars, and two digits to the right of the decimal place were sufficient. Business-oriented computers could therefore use a shorter word length or a variable word length, if there was a way to tell the processor when the end of a word was reached. The IBM 702, IBM 1401, RCA 301, and RCA 501 had variable word lengths, with the end of a word set by a variety of means. The 1401 used an extra bit appended to each coded character to indicate whether or not it was the last one of a word; the 702 used a a special character that signified that the end was reached.²⁷ Although popular in the 1950s, computers with variable word lengths fell from favor in the following decades and are no longer common.

用于非数字应用（尤其是用于控制其他设备的计算机）也可以使用较短的字长。其中最有影响力的是 Whirlwind，其字长为 16 位。利用短字长优势的商用机器于 1959 年左右开始出现，其中包括控制数据公司的 CDC 160（12 位）和数字设备公司的 PDP-1（18 位）。

Computers intended for non-numeric applications, especially for controlling other devices, could also use a shorter word length. The most influential of these was the Whirlwind, which had a word length of 16 bits. Commercial machines that exploited the advantages of a short word length began to appear around 1959, and included the Control Data Corporation CDC 160 (12 bits) and the Digital Equipment Corporation PDP-1 (18 bits).

寄存器结构早期计算机的处理单元包含一组电路，这些电路可以保存数值并对其进行基本算术运算 — 通常只是简单的加法。这种设备被称为累加器，因为可以在其中建立或“累积”总和；具有更通用但也是临时存储能力的设备称为寄存器（图2.4）。另一组电路组成了程序计数器，它存储了处理器要从内存中获取并执行的程序指令的位置。

Register Structure Processing units of early computers contained a set of circuits that could hold a numeric value and perform rudimentary arithmetic on it—usually nothing more than simple addition. This device became known as an accumulator, since sums could be built up or “accumulated” in it; a device with more general, though also temporary, storage ability is called a register (figure 2.4). Another set of circuits made up the program counter, which stored the location of the program instruction that the processor was to fetch from memory and execute.

处理器的典型周期是从内存中获取一条指令，对累加器中的数据执行该指令，并使用下一条指令的地址更新程序计数器。在最简单的情况下，程序计数器会自动加一（因此称为“计数器”），但分支指令可以指定计数器指向不同的内存位置。

The typical cycle of a processor was to fetch an instruction from memory, carry out that instruction on data in the accumulator, and update the program counter with the address of the next instruction. In the simplest case, the program counter was automatically incremented by one (hence the term “counter”), but branch instructions could specify that the counter point to a different memory location.

计算机程序命令处理器执行算术或逻辑运算（例如，添加或比较），告诉处理器相关数据的位置，并告诉它在哪里存储结果。与顺序获取指令一样，处理器通常需要将数据存储在内存中彼此靠近的位置。因此，对此类数据列表执行相同操作的程序可能由一长串相同操作组成，后面只是地址略有变化。或者程序可以修改自身并在每次执行操作时更改地址。（这个想法可能源自冯·诺依曼在他早期的计算机设计报告中。）这两种过程都不太优雅。

A computer program orders the processor to perform arithmetic or logic (e.g., add or compare), tells the processor where the relevant data are to be found, and tells it where to store results. As with the sequential fetching of instructions, often the processor requires data that are stored close to one another in memory. A program that performs the same operation on such a list of data might therefore consist of a long list of the same operation, followed by only a slight change in the address. Or the program could modify itself and change the address each time an operation is executed. (This idea may have originated with von Neumann in his early reports on computer design.) Neither process is elegant.

1948 年，设计师们开始在曼彻斯特大学设计一台实验计算机，并在处理器中增加了一个额外的索引寄存器，以简化数据数组的处理。（在早期出版的曼彻斯特计算机描述中，其设计者称该寄存器为“B 线”，并使用符号“A”表示“累加器”，“C”表示“控制”。这些术语一直沿用到 20 世纪 50 年代的处理器描述中。）通过指定一个值来增加指令的地址字段，程序不再需要像冯·诺依曼和其他先驱者设想的那样修改自身。这大大简化了本来就很困难的编程过程^。28

Designers began with an experimental computer at the University of Manchester in 1948, and added to the processor an extra index register to simplify working with arrays of data. (In early published descriptions of the Manchester computer, its designers called this register a “B-line,” and used the symbol “A” for “accumulator,” and “C” for “control.” These terms persisted into the 1950s in descriptions of processors.) By specifying a value to increment the address field of an instruction, programs no longer had to modify themselves as envisioned by von Neumann and other pioneers. That greatly simplified the already difficult process of programming.²⁸

图 2.4

计算机中央处理器 (CPU)，显示不同的寄存器。标有“乘除”的寄存器通常具有其他寄存器的两倍字长，因为一般来说，两个n位数字相乘会产生具有 2 n位数字的乘积。下图：单地址指令由一侧的操作字段和另一侧的地址字段组成。中间是一组执行特殊功能的位，例如指定哪个索引寄存器保存所需的数字。

Figure 2.4

Computer central processing unit, or CPU, showing the different registers. The register labeled “Multiply–Divide” typically had twice the word length of the others, because, in general, a multiplication of two n-digit numbers gives a product having 2n digits. Below: a single-address instruction consisted of an operation field on one side, and an address field on the other. In between was a set of bits that performed special functions, such as specifying which of the index registers held a desired number.

这三种类型的寄存器（累加器、程序计数器和 B 行或索引寄存器）构成了 20 世纪 50 年代大多数大型计算机的处理单元。例如，1954 年发布的 IBM 704 具有 36 位字长、可容纳 4,096 个字的核心内存以及带有累加器、程序计数器和三个索引寄存器的处理器。另一个寄存器与累加器耦合，专用于乘法和除法（例如，用于存储两个 36 位数相乘时生成的额外位）^。29

These three types of registers—accumulator, program counter, and B-line or index register—made up the processing units of most large computers of the 1950s. For example, the IBM 704, announced in 1954, had a 36-bit word length, a core memory holding 4,096 words, and a processor with an accumulator, program counter, and three index registers. Another register was coupled to the accumulator and dedicated to multiplication and division (e.g., to store the extra bits that are generated when two 36-bit numbers are multiplied together).²⁹

1956 年，英国公司 Ferranti Ltd. 宣布推出一款名为 Pegasus 的机器，其处理器包含一组八个寄存器，其中七个可用作累加器或索引寄存器。这开创了提供通用寄存器的概念，程序可以根据特定程序的需要将其用于上述任何功能。其他公司采用这种理念的速度很慢，但到下一个十年末，它成为最受欢迎的设计^。30

In 1956 the British firm Ferranti Ltd. announced a machine, called Pegasus, whose processor contained a set of eight registers, seven of which could be used as accumulators or as index registers. That inaugurated the notion of providing general-purpose registers that a program could use for any of the above functions, as needed by a specific program. Other companies were slow to adopt this philosophy, but by the end of the next decade it became the most favored design.³⁰

地址数基于累加器的机器的指令分为两部分：第一部分指定操作（例如，加、减或比较），第二部分指定要操作的数据的地址。如果一个操作需要两部分数据，则另一个操作数需要存在于累加器中。它可以是前一个操作的结果，也可以是将其从内存加载到累加器的明确指令的结果。由于每条指令都包含一个内存引用，因此这称为单地址方案。它的优点是简单；它还允许指令的地址字段足够长以指定大部分内存。20 世纪 50 年代制造的许多计算机都使用了它，包括最初的 UNIVAC 和 IBM 的大型科学计算机系列 701、704、709 和 7090。

Number of addresses Instructions for an accumulator-based machine had two parts: The first specified the operation (e.g., add, subtract, or compare) and the second the address of the data to be operated on. If an operation required two pieces of data, the other operand needed to be present in the accumulator. It could be there as the result of the previous operation, or as a result of an explicit instruction to load it into the accumulator from memory. Because each instruction contained one memory reference, this was called a single-address scheme. Its virtue was its simplicity; it also allowed the address field of an instruction to be long enough to specify large portions of memory. Many computers built in the 1950s used it, including the original UNIVAC and IBM’s series of large scientific computers, the 701, 704, 709, and 7090.

单地址方案还有其他替代方案。一种方案是让一个操作后跟两个地址，用于两个操作数。可以添加第三个地址字段，将操作结果存储在内存中，而不是假设它们会保留在累加器中。UNIVAC 1103、RCA 601 和 IBM 1401 使用双地址方案，而 UNIVAC 文件计算机和 Honeywell H-800 使用三地址方案。

There were alternatives to the single-address scheme. One was to have an operation followed by two addresses, for both operands. A third address field could be added, to store the results of an operation in memory rather than assume they would remain in the accumulator. The UNIVAC 1103, RCA 601, and IBM 1401 used a two-address scheme, while the UNIVAC File Computer and Honeywell H-800 used a three-address scheme.

这些方案都有地址字段，用于指示数据的位置。还可以包括下一条指令的地址，而不必转到程序计数器。IBM 650 和 Librascope LGP-30 等鼓式计算机使用这种方法来最大限度地减少在鼓上搜索下一条指令所花费的时间——每条指令都可以在执行上一条指令后将计算机引导到鼓上所需数据所在的位置。这种方式编程很困难，但它解决了鼓式机械固有的慢速问题。随着磁芯的出现，这种方案不再必要^。31

These schemes all had address fields that told where data were located. One could also include the address of the next instruction, instead of going to the program counter for it. Drum computers like the IBM 650 and Librascope LGP-30 used this to minimize the time spent searching the drum for the next instruction—each instruction could direct the computer to the place on the drum where the desired data would be, after executing the previous instruction. Programming this way was difficult, but it got around the inherently slow speeds of drum machinery. With the advent of the magnetic core, this scheme fell from necessity.³¹

最后，可以设计一条根本不指定地址的指令：两个操作数始终以正确的顺序保存在指定的一组寄存器中。结果同样以正确的顺序进入该位置以供进一步处理。这需要将一组寄存器（或内存位置）组织成一个称为堆栈的结构，该结构以后进先出 (LIFO) 的方式向处理器呈现数据。（LIFO 的概念来自会计行业；请注意它与自助餐厅里弹簧式盘子堆的相似性。）³²采用这种方案的计算机最早出现在 20 世纪 60 年代，但它们从未真正挑战过单地址设计。堆栈架构在 20 世纪 70 年代中期在可编程袖珍计算器中重新出现。

Finally, one could design an instruction that specified no addresses at all: both operands were always kept in a specified set of registers, in the correct order. Results likewise went into that place, in the proper order for further processing. That required organizing a set of registers (or memory locations) in a structure called a stack, which presented data to the processor as Last-In, First-Out (LIFO). (The concept of LIFO came from the accounting profession; note its similarity to the spring-loaded stacks of plates found in cafeterias.)³² Computers with this scheme first appeared in the 1960s, but they never seriously challenged the single-address design. Stack architecture resurfaced in the mid-1970s in programmable pocket calculators.

I/O 通道和“轮回之轮” UNIVAC 的创新之一是使用存储区域作为慢速输入和输出设备（如读卡器和电动打字机）与速度更快的中央处理器之间的“缓冲区”。同样，UNIVAC 1103A 引入了“中断”的概念，使机器的处理器能够处理问题，仅在必要时停止处理与外界之间的数据传输。这些创新逐渐成熟，并扩展到 20 世纪 50 年代的大型商用机器。

I/O Channels and the “Wheel of Reincarnation” One of the UNIVAC’s innovations was its use of a storage area that served as a “buffer” between the slow input and output equipment, such as card readers and electric typewriters, and the much faster central processor. Likewise the UNIVAC 1103A introduced the concept of the “interrupt,” which allowed the machine’s processor to work on a problem, stopping to handle the transfer of data to or from the outside world only when necessary. These innovations became well-established and were extended to the large commercial machines of the 1950s.

随着 I/O 速度与中央处理器匹配的要求变得越来越复杂，处理这种传输的设备也变得越来越复杂。随着 IBM 709 的推出，IBM 工程师设计了一个单独的处理器，称为“通道”，来处理输入和输出。这个通道实际上是一台小型计算机，专门用于管理以不同数据速率运行的各种 I/O 设备的特定问题。33^有时，当设计师推出改进的计算机模型时，他们会增加这个通道的功能，直到它与主处理器一样复杂和强大——现在他们有一个双处理器系统。那时，冯·诺依曼架构的优雅简洁性有被遗忘的危险。为了重新获得它，必须重新考虑计算机对处理和 I/O 的基本要求。这个所谓的轮回之轮，如果不被新的设计方法打破，就可能导致一个巴洛克式和笨重的系统^。34

As the requirements for matching the speeds of I/O with the central processor grew more complex, so too did the devices designed to handle this transfer. With the introduction of the IBM 709, IBM engineers designed a separate processor, called a “channel,” to handle input and output. This channel was, in effect, a small computer dedicated to the specific problem of managing a variety of I/O equipment that operated at different data rates.³³ Sometimes, as designers brought out improved models of a computer, they would add to the capabilities of this channel until it was as complex and powerful as the main processor—and they now had a two-processor system. At that point the elegant simplicity of the von Neumann architecture was in danger of being lost. To recapture it, the computer’s basic requirements for processing and I/O had to be reconsidered. This so-called wheel of reincarnation, if not broken by a fresh design approach, threatened to lead to a baroque and cumbersome system.³⁴

I/O 通道的复杂性推高了系统成本，但对于使用计算机处理大量数据问题的客户来说，通道是必不可少的。随着时间的推移，通道成为大型计算机的一个决定性特征——价格昂贵、体积庞大，并且具有足够的内存、灵活性、I/O 设施和处理速度来满足大型客户的需求。大型计算机成为 20 世纪 60 年代的标准，尽管其他类型的计算机也出现了，既有高端计算机（需要更快的处理速度和更简单的 I/O），也有低端计算机（主要设计目标是低成本）。

The complexity of I/O channels drove up the cost of systems, but they were necessary for customers who used computers for problems that handled large quantities of data. In time, channels became a defining characteristic of the mainframe computer—one that was expensive, physically large, and contained enough memory, flexibility, I/O facilities, and processing speed to handle the needs of large customers. The mainframe computer became the standard of the 1960s, although other classes would arise both at the higher end, where faster processing but simpler I/O was required, and at the lower end, where low cost was a major design goal.

浮点硬件最后需要提到的一个设计特点是，它主要关注科学应用，但也对商业客户产生了影响。用一位计算机设计师的话来说，一台机器是否在其硬件中处理浮点运算是“区分小型计算机和大型计算机的最大且可能唯一的因素”^。35

Floating-point Hardware One final design feature needs to be mentioned, which was of concern primarily to scientific applications, but had an impact on commercial customers as well. In the words of one computer designer, whether or not a machine handles floating-point arithmetic in its hardware is the “biggest and perhaps only factor that separates a small computer from a large computer.”³⁵

浮点运算允许用户跟踪计算的总体规模。它通过将一个量分成两部分来实现这一点，其中一部分用于标记小数点（机器内部的二进制小数点）的位置。例如，化学家经常使用一个称为阿伏伽德罗常数的量，他们将其写为 6.02 × 10 ²³。以科学计数法表示时，该数字分为两部分，第二部分表示幅度。在此示例中，23 表示小数点位于数字第一部分右侧 23 位。如果以普通计数法表示，阿伏伽德罗常数必须写为 602 后面跟着 21 个零。科学家和工程师对它很熟悉（有时称之为“科学记数法”），但在商业世界中几乎不为人知，因为商业计算的最高限度不超过数万亿美元，最低限度也不超过万亿分之一美元。（自 1980 年以来，通货膨胀使数百万甚至数十亿美元的数字变得更加普遍。与此同时，它降低了美分的价值；现在人们甚至懒得在街角商店买一分钱。虽然税率仍以密尔（十分之一美分）为单位，但不太可能有人会像亨利·戴维·梭罗那样，精确到半便士地计算房屋的价值。）

Floating-point arithmetic allows users to keep track of the overall scale of a computation. It does so by dividing a quantity into two parts, one of which serves to mark the place of the decimal point (binary point inside the machine). For example, chemists often work with a quantity known as Avogadro’s number, which they write as 6.02 × 10²³ . When written in scientific notation, the number has two parts, the second of which indicates the magnitude. In this example, the 23 indicates that the decimal point belongs 23 places to the right of where it appears in the first part of the number. If written in ordinary notation, Avogadro’s number would have to be written as 602 followed by 21 zeroes. It is familiar to scientists and engineers (sometimes under the term “scientific notation”), but almost unknown in the commercial world because commercial calculations do not reach beyond trillions of dollars at the upper end, nor to trillionths of a dollar at the low end. (Inflation has made million-and even billion-dollar figures a lot more common since 1980. At the same time, it has diminished the value of cents; people now do not even bother with pennies at the corner store. Although tax rates are still given in mills (one tenth of a cent), it is unlikely that anyone will calculate the value of a house, as Henry David Thoreau did, to the halfpenny.)

任何计算机都可以编程为浮点计算机，但编程很复杂，会降低计算机的速度。另一方面，如果计算机的电子电路设计为在硬件中处理浮点，则处理器将更加复杂和昂贵。在 20 世纪 40 年代后期，人们争论哪种方法更好，但应该指出的是，最早的机电计算机（由德国的 Konrad Zuse 和美国贝尔实验室在 20 世纪 40 年代制造的）已经内置了浮点。

Any computer can be programmed to operate as a floating-point machine, but the programming is complex and slows the machine down. On the other hand, if the electronic circuits of a machine are designed to handle floating point in hardware, the processor will be more complicated and expensive. In the late 1940s there were debates over which approach was better, although it should be noted that the very first electromechanical computers, those built by Konrad Zuse in Germany and by Bell Labs in the United States in the 1940s, had floating point wired in.

制造商认为商业客户不需要浮点数，也不愿意为此付费。他们通常提供两条独立的机器生产线，一条用于商业客户，另一条用于科学或工程应用。前者包括 UNIVAC、UNIVAC File 和 IBM 702 和 705，只有定点算术，而且通常字长可变。后者，如 704 和 709，有浮点数和相对较长的固定字长。面向商业的机器的 I/O 设施通常更复杂。

Manufacturers felt that commercial customers did not need floating point and would be unwilling to pay for it. They typically offered two separate lines of machines, one for commercial customers and the other for scientific or engineering applications. The former, which included the UNIVAC, the UNIVAC File, and IBM 702 and 705, had only fixed-point arithmetic, and often a variable word length. The latter, like the 704 and the 709, had floating point and a relatively long, fixed word length. I/O facilities were often more elaborate for the business-oriented machines.

这两个几乎相似的设备系列的并行开发一直持续到 20 世纪 50 年代末，直到下一个十年。对于许多客户来说，两者的区别并不那么明显。例如，面向商业客户的 IBM 650 通常安装在大学中心，教授和学生在那里努力为其开发浮点软件。同样，IBM 704 的性能优于 705，因为许多人认为它的架构更优越，订购 704 用于科学工作的客户很快发现自己也将其用于商业应用。随着编程语言 FORTRAN 可以在 704 上使用，人们对 704 的偏好进一步增加。IBM 于 1964 年推出了 System/360 系列计算机，将这两条产品线合并在一起。

This parallel development of two almost-similar lines of equipment persisted through the late 1950s into the next decade. For many customers the distinction was not that clear. For example, the IBM 650, intended for commercial customers, was often installed in university centers, where professors and students labored to develop floating-point software for it. Likewise, the IBM 704 had better performance than the 705, because of what many felt was a superior architecture, and customers who ordered a 704 for scientific work soon found themselves using it for commercial applications as well. The preference for the 704 increased even more as the programming language FORTRAN became available on it. IBM combined both lines with the introduction of the System/360 line of computers in 1964.

晶体管

The Transistor

尽管早在 20 世纪 50 年代初，用于替代真空管和磁芯存储器的晶体管就在实验室中投入使用，但直到 20 世纪 50 年代末，也就是磁芯普遍使用之后，晶体管的可靠性才足以用于商用计算机。

Although the transistor as a replacement for the vacuum tube and core memory were both working in the laboratory by the early 1950s, the transistor was not reliable enough to be used in commercial computers until late in that decade, well after cores were in common use.

发明晶体管的贝尔实验室并不是 20 世纪 50 年代考虑进入商用计算机市场的公司之一。该公司是一家受监管的垄断企业，在采取每一项行动时都会考虑联邦法院的裁决。1956 年初，经过 7 年的诉讼，该公司通过一项同意令解决了美国司法部提起的诉讼，同意不从事“除提供公共承运人通信之外”的任何业务。36 ^AT &T 在战后繁荣的郊区安装电话的业务已经足够多了。如果它不使用晶体管制造计算机，它可以使用它们来取代交换设备和电话放大器中的真空管。

Bell Laboratories, where the transistor was invented, was not among those considering entering the commercial computer market in the 1950s. The company was a regulated monopoly and weighed every action it took with an eye on the federal courts. In early 1956, after seven years of litigation, it settled a lawsuit brought by the U.S. Justice Department with a consent decree in which it agreed not to enter into any business “other than the furnishing of common carrier communications.”³⁶ AT&T had enough business installing telephones in the booming postwar suburbs anyway. If it did not use transistors for computers, it could use them to replace vacuum tubes in switching equipment and in telephone amplifiers.

AT&T，包括贝尔实验室及其西部电气制造部门，当时也拥有大量军事业务。贝尔实验室在 1952 年左右为军方制造了几台专用计算机，包括一台数字数据传输装置和一台名为 TRADIC 的专用计算机。37^在整个 20 世纪 50 年代和 60 年代，他们为空军和陆军导弹系统提供计算设备，包括 NIKE 和 Titan。这为其他公司奠定了基础，经过十年的发展，这些公司终于开始供应使用晶体管的商用计算机。

AT&T, including Bell Laboratories and its Western Electric manufacturing arm, also had a substantial military business at the time. Bell Labs built several special-purpose computers for the military around 1952, including a digital data transmission set, and a special-purpose computer called TRADIC.³⁷ Throughout the 1950s and 1960s they provided computing equipment for Air Force and Army missile systems, including the NIKE and Titan. This laid the foundation for other companies, who after a decade of development finally began to supply commercial computers using transistors.

飞歌

Philco

部分是为了满足联邦监管机构的要求，贝尔实验室以象征性的价格公开了晶体管的信息。在众多开始生产晶体管的公司中，有一家总部位于费城的老牌电子公司 Philco。几年之内，Philco 就率先开发出一种晶体管，他们称之为“表面屏障”，这种晶体管可以大批量生产，性能良好。1954 年，Philco 的晶体管被用在麻省理工学院的 TX-0 实验计算机中，后来又被用在 Philco 自己的机器中^。38

In part to satisfy federal regulators, Bell Labs made information about transistors available at a nominal cost. Among the many companies that began producing transistors was Philco, an established electronics firm headquartered in Philadelphia. Within a few years Philco pioneered a type of transistor, which they called “surface barrier,” that could be made in quantity and that had good performance. Philco’s transistors were used in the TX-0 experimental computer at MIT in 1954, and later in Philco’s own machines.³⁸

Philco 在生产表面势垒晶体管方面处于领先地位，这使该公司一跃成为计算领域的领军企业。1955 年 6 月，该公司与美国国家安全局签订合同，生产基于 UNIVAC 1103（本身是为该机构制造的计算机的商业版本）架构的快速计算机。39^最终成果被称为“SOLO”，于 1956 年至 1958 年之间完成，可能是美国第一台运行的通用晶体管计算机。40 ^Philco推出了一款名为 TRANSAC S-1000 的商业版本，随后在 1958 年推出了升级版 S-2000。S-2000 的首批交付时间为 1960 年 1 月。这些产品以及 UNIVAC 交付的名为 Solid State 80 的小型计算机标志着晶体管时代或“第二代”的开始^。41

Philco’s lead in producing surface-barrier transistors catapulted the company to the forefront of computing. In June 1955 it contracted with the National Security Agency to produce a fast computer based on the architecture of the UNIVAC 1103 (itself a commercial version of a computer built for that agency).³⁹ The result, called “SOLO,” was completed sometime between 1956 and 1958, and was probably the first general-purpose transistorized computer to operate in the United States.⁴⁰ Philco marketed a commercial version called the TRANSAC S-1000, followed quickly by an upgraded S-2000, in 1958. First deliveries of the S-2000 were in January 1960. These, along with deliveries by UNIVAC of a smaller computer called the Solid State 80, mark the beginning of the transistor era, or “Second Generation.”⁴¹

S-2000 是一款大型、昂贵的机器，但在美国国防部等客户中销量很好，因为他们总是要求顶级性能。包括通用电气和联合航空在内的几家商业客户都购买了 S-2000，以替代他们已经在使用的大型 IBM 系统。42^在建立滩头阵地后，Philco 发现自己面临着不断创新和升级产品的巨大压力。这需要资金和人力，而公司不愿或无法提供。1962 年，在被福特汽车公司收购后不久，Philco 退出了计算机业务。

The S-2000, a large, expensive machine, sold well to those customers, especially in the U.S. Defense Department, who always demanded topranking performance. Several commercial customers, including GE and United Aircraft, bought one as a replacement for the large IBM systems they were already using.⁴² Having established a beachhead, Philco found itself under strong pressure to constantly innovate and upgrade its products. That demanded money and manpower the company was unwilling, or unable, to provide. In 1962, shortly after being acquired by Ford Motor Company, Philco dropped out of the computer business.

NCR，巴勒斯

NCR, Burroughs

晶体管的发展也影响了 National Cash Register 的计算机业务，该公司于 1954 年通过收购计算机研究公司 (CRC) 进入计算机行业，CRC 是由一群来自 Northrop Aircraft 的工程师创立的公司。尽管 NCR 在二战期间制造了专门的设备来协助美国军方破译敌方密码，但这并没有使其偏离战前对机械收银机的关注。1954 年至 1957 年间，该公司并没有积极推销受 CRC 启发的电子计算机，当 NCR 在 1957 年发布大型晶体管计算机 304 型时，它将生产和制造移交给了 GE。304 型的适度销售（最终安装了 33 台）为 NCR 后来的成功做出了贡献（并阻止了 GE 完全退出计算机领域）。43^由于专注于零售、银行业务和其他一些专业但规模大且利润丰厚的商业利基市场，该公司并非一直繁荣，但它确实度过了从黄铜收银机时代到电子收银机时代的痛苦过渡期。 AT&T 于 1991 年收购了该公司，试图成为商业计算领域的竞争对手。AT&T 失败了，并于 1995 年将 NCR 拆分为一家独立公司。

The development of transistors also affected the computer business of National Cash Register, which entered the computer industry in 1954 by buying the Computer Research Corporation (CRC), a company founded by a group of engineers from Northrop Aircraft. Although during World War II NCR had built specialized equipment to assist the U.S. military in breaking enemy codes, that had not led it away from its prewar focus on mechanical cash registers. Between 1954 and 1957 the company did not market CRC-inspired electronic computers aggressively, and when in 1957 NCR announced the model 304, a large transistorized computer, it turned over production and manufacturing to GE. Modest sales of the 304 (eventually thirty-three were installed) contributed to NCR’s later success (and kept GE from leaving the computer field entirely).⁴³ With its focus on retail sales, banking, and a few other specialized, but large and profitable, commercial niches, the company did not always prosper, but it did survive a painful transition from the era of brass cash registers to one of electronics. AT&T bought the company in 1991, in an attempt to become a competitor in commercial computing. AT&T failed, and it spun off NCR as an independent company in 1995.

宝来公司是一家加法机和银行设备制造商，与 NCR 一样进入了晶体管时代。1956 年，宝来收购了一家生产基于鼓的科学计算机的小公司 Electrodata，该公司是帕萨迪纳联合工程公司的一个部门。Electrodata 的工程师与喷气推进实验室关系密切，他们制造了一台计算机，与其他基于鼓的计算机一样，速度很慢，但结构设计却很出色。它是美国制造的第一批在处理器中配备索引寄存器的计算机之一，⁴⁴这使它成为知识渊博的客户（尤其是大学）的最爱。

The Burroughs Corporation, a manufacturer of adding machines and banking equipment, entered the transistorized era like NCR. In 1956 Burroughs purchased a small firm that made a drum-based scientific computer, Electrodata, a division of Consolidated Engineering of Pasadena. Staffed by engineers who had close ties to the Jet Propulsion Laboratory, Electrodata had built a computer that was slow, like other drum-based computers, but had a superior architectural design. It was among the first American-made computers to have index registers in its processor, ⁴⁴ which made it a favorite among the more knowledgeable customers, particularly in universities.

Burroughs 当时已经推出了一款由其内部设计的机器 E-101，该公司认为这款机器更适合购买其会计机器的典型商业客户。E-101 只比办公桌大一点，而且价格便宜，售价约为 35,000 美元。它缺乏存储程序架构。45^它的销量不佳。市场再次选择了设计精良的存储程序计算机，而不是针对特定需求的较便宜的机器。

Burroughs was already offering a machine it designed in-house, the E–101, which it felt was better suited for the typical commercial customers that bought its accounting machines. The E-101 was only a little bigger than a desk, and inexpensive, selling for about $35,000. It lacked a stored-program architecture.⁴⁵ It did not sell very well. Once again, the marketplace chose a well-designed stored-program computer over a less expensive machine intended for specific needs.

Burroughs 的 E-101 失败，加上未能及时将卓越的 Electrodata 设计应用于晶体管，导致其在商业市场的影响力很小。它通过军事合同保持了其计算机专业知识，其中包括为 SAGE 防空系统制造专用计算机的大型合同。另一份合同是 Atlas 洲际弹道导弹的制导系统，它导致了一台晶体管计算机的出现，该计算机在 1957 年至 1959 年间成功引导了 Atlas 的发射。Burroughs 声称这是“第一台使用晶体管而不是真空管的可操作计算机”。46^上面描述的 SOLO 计算机可能早于它，但 SOLO 的存在多年来一直保密。那段经历为 Burroughs 在 20 世纪 60 年代成功重返商业计算领域奠定了基础。

Burroughs’s failure with the E-101, combined with its failure to adapt the superior Electrodata design to transistors quickly enough, kept Burroughs’s influence in the commercial market small. It was able to keep its computer expertise intact through military contracts, including a large contract to build specialized computers for the SAGE air defense system. Another contract, for the guidance system for the Atlas intercontinental ballistic missile, led to a transistorized computer that successfully guided Atlas launches from 1957 through 1959. Burroughs claimed that this was “the first operational computer to use transistors rather than vacuum tubes.”⁴⁶ The SOLO computer, described above, may have preceeded it, but SOLO’s existence was kept secret for many years. That experience laid a foundation for Burroughs’s successful reentry into commercial computing in the 1960s.

IBM 的崛起

The Rise of IBM

到 1960 年，IBM 已占据计算机行业主导地位。随之而来的是人们对其股票的浓厚兴趣，这又促使大批金融和技术分析师关注该公司并试图预测其下一步行动。IBM 始终对其公众形象十分谨慎，这在一定程度上源于 Thomas Watson Sr. 所感受到的羞辱，他觉得 1912 年在 National Cash Register 工作时因违反反垄断法而被判有罪是不公平和毫无根据的。47 ^IBM的公关部门始终礼貌而及时地向外界提供信息，但这些信息始终经过精心组织以反映形象。

By 1960 IBM dominated the computer industry. With that came an intense interest in its stock, which in turn gave rise to a legion of financial and technical analysts who watched the company and tried to discern its next move. IBM was always careful of its public image, in part a legacy of the humiliation felt by Thomas Watson Sr. over what he felt was an unfair and unwarranted conviction for violating antitrust laws while he was working for National Cash Register in 1912.⁴⁷ IBM’s public relations department always courteously and promptly supplied outsiders with information, but the information was always carefully structured to reflect an image.

这反过来又导致一群人开始揭穿、贬低或以其他方式拆散该公司。当然，他们当中有 IBM 的主要竞争对手，但也包括来自大学的计算机专业人士和美国司法部的官员。1952 年，在 IBM 进入电子计算机业务之前，司法部指控其在开展打孔卡业务时违反了反垄断法。这导致了 1956 年的一项同意令，该令并没有解散 IBM，但却深深影响了 IBM 和计算机行业。也许其中最重要的规定是 IBM 允许客户以与其向租户提供的条件相当的条件购买其机器。这标志着 IBM 的重大转变，并最终导致一个庞大的子行业，即购买 IBM 大型机并将其出租给其他客户^。48

That, in turn, led to groups of people who set out to debunk, disparage, or otherwise dismantle the company. Among them were, of course, IBM’s main competitors, but they also included both computer professionals from universities and officials in the U.S. Justice Department. In 1952, before IBM entered the electronic computer business, the Justice Department had alleged that it violated anti-trust laws in conducting its punched card business. That led, in 1956, to a Consent Decree, which did not break IBM up but deeply affected it and the computer industry. Perhaps its most important provision was that IBM was to allow customers to purchase its machines on terms comparable to those it offered to renters. That signaled a major shift for IBM, and would eventually lead to a large sub-industry of companies that purchased IBM mainframes and leased them out to other customers.⁴⁸

随着 IBM 股价飙升，许多在几十年前购买了股票的员工成为百万富翁，“IBM 学家”对这家公司既着迷又厌恶，他们开始寻找公司内部运作的线索。就像“克里姆林宫学家”测量苏联官方照片中各官员的身高一样，IBM 观察家们从他们能找到的一切（新闻剪报、产品公告、财务报告中隐藏的数字）中寻找模式。49 这在《Datamation》上表现^得最为明显，这是一本 1957 年创办的行业杂志，原名为《研究与工程》，面向计算机行业。该杂志的所有者最初打算出于税收目的将其注销，但他们很快意识到，《Datamation》的主题越来越受关注，也是广告收入的一个很好的来源。⁵⁰在 Santo (Sandi) Lanzarotta 的编辑指导下，尤其是 1963 年之后的 Robert Forest 的指导下，Datamation形成了完美的平衡：对 IBM 所谓的高压手段进行无情的批评，同时又对计算及其对社会的益处充满热情，大多数读者都认识到 IBM 也抱有这种热情，并以自己的方式进一步发扬光大。

As IBM’s stock soared and made millionaires of many employees who had bought shares in earlier decades, “IBM-ologists,” at once fascinated and repelled by the company, searched for clues to its inner workings. Like “Kremlin-ologists,” who measured how tall various officials appeared in official Soviet photographs, IBM-watchers combed whatever they could find—press clippings, product announcements, figures buried within financial reports—for patterns.⁴⁹ Nowhere was this more evident than in the pages of Datamation, a trade journal founded as Research & Engineering and oriented toward the computer industry in 1957. The magazine’s owners at first intended to write it off for tax purposes, but they soon recognized that Datamation’s subject was one of growing interest, and also a very good source of advertising revenue.⁵⁰ Under the editorial hand of Santo (Sandi) Lanzarotta, and especially Robert Forest after 1963, Datamation developed a perfect balance: relentless criticism of IBM for its alleged heavy hand, tempered by a passion for computing and its benefits to society, a passion that most readers recognized was one that IBM shared and furthered in its own way.

批评者指责 IBM 从来不是创新者，总是等到另一家较小的公司承担技术风险，然后通过可疑的营销手段一举占领市场。至少与 Eckert 和 Mauchly 相比，IBM 很晚才认识到电子计算的未来。有些人认为 IBM 701 的 I/O 设计不如 UNIVAC。另一方面，IBM 704 拥有浮点运算、FORTRAN 编程语言和核心内存，在技术上优于 UNIVAC 在 1956 年提供的产品，而那时最初的 UNIVAC 已经过时了。704 的销售是 IBM 崛起的主要因素，而这些销售并非完全是公司营销和销售团队的结果。

Critics charged that IBM was never an innovator but always waited until another, smaller company took the technical risks, and then swept in and took over by questionable marketing practices. IBM was late in recognizing the future of electronic computing, at least compared to Eckert and Mauchly. And some regarded the IBM 701 as inferior to the UNIVAC because of its I/O design. On the other hand, the IBM 704, with its floating-point arithmetic, FORTRAN programming language, and core memory, was technically superior to what UNIVAC offered in 1956, by which time the original UNIVAC was then obsolete. Sales of the 704 were a major factor in IBM’s ascendancy, and those sales were not entirely the result of the company’s marketing and sales force.

仔细研究这一指控就会发现，IBM 还通过卓越的内部制造技术和现场服务弥补了其在技术创新方面的落后。这些努力融入了营销和销售，这总是很激进，竞争对手常常觉得不公平。然而，对于 IBM 来说，并不总是能够将纯粹的技术层面与销售技巧区分开来。以下两个例子可以说明营销、制造和技术创新的这种融合。

A closer look at this charge reveals that IBM also made up for its lag in technical innovation by superior in-house manufacturing techniques and field service. These efforts merged into marketing and sales, which was always aggressive and which competitors often felt was unfair. However, with IBM it is not always possible to separate the purely technical dimension from the salesmanship. The following two examples may serve to illustrate this merging of marketing, manufacturing, and technical innovation.

磁盘存储1957 年，IBM 推出了一款后来被证明是其最持久的设备：用于随机访问大量数据的旋转磁盘。磁盘驱动器是自本世纪初以来一直是电子计算机主流的磁鼓存储器的近亲，但它的几何形状使更多的表面积暴露于磁化和存储。IBM 完善了一项源于 20 世纪 40 年代中期 Presper Eckert 的创新，并将其推向市场，当时他建议在 ENIAC 的继任者中使用磁盘来存储程序和数据。1952 年，美国国家标准局根据马里兰州阿伯丁弹道研究实验室的订单，开始研发磁盘存储器^。51

Disk Storage In 1957 IBM marketed a device that would prove to be one of its most lasting: the spinning disk for random-access storage of large quantities of data. The disk drive was a cousin to the drum store that had been a mainstay of electronic computers since the beginning of the decade, but it had a geometry that exposed more surface area to magnetization and storage. IBM perfected and brought to market an innovation that had originated in the mid-1940s with Presper Eckert, who had suggested using a disk for program and data storage for the ENIAC’s successor. And in 1952 the National Bureau of Standards was working on a disk store in response to an order by the Ballistic Research Laboratory at Aberdeen, Maryland.⁵¹

与磁鼓相比，使用旋转磁盘阵列可以提供更大的容量，并且每位成本更低，但也存在棘手的技术问题。典型的磁鼓存储器有一组固定的磁头，它们沿着一条线牢固地固定，磁鼓上的每个轨道都有一个磁头。整个机制可以做得坚固耐用，但成本高昂。磁盘阵列永远不可能像磁鼓那样坚固，而且似乎访问磁盘表面的唯一方法是让磁头像唱机的唱针一样放置在磁盘上方。要记录和读取数据，磁头必须非常靠近表面。如果它们接触（就像磁带驱动器的磁头接触磁带一样），旋转磁盘的高速度可能会造成不可接受的磨损。

Using an array of spinning disks offered much greater capacity at lower cost per bit than drums, but it also presented knotty technical problems. Typical drum stores had a set of fixed heads rigidly fastened along a line, one for each track on the drum. The whole mechanism could be made rugged and stable, but it was expensive. A disk array could never be made as rigid as a drum, and it seemed the only way to access the surfaces of the disks was to have heads that could be positioned over the disk like the stylus of a record player. To record and read data, the heads had to be very close to the surfaces. If they touched (as the head of a tape drive touches the tape), the high speeds of the spinning disks would probably cause unacceptable wear.

在加利福尼亚州圣何塞新成立的实验室工作的 IBM 工程师提出了使用薄薄的空气膜作为缓冲垫以防止磁头接触磁盘表面的想法。他们的第一款产品使用外部压缩机向磁头提供的空气。后来，IBM 磁盘驱动器巧妙地利用了“边界层”——一种飞机设计师熟悉的空气动力学现象——通过磁盘本身产生的空气使磁盘漂浮起来。实验室在 1953 年和 1954 年试验了各种令人眼花缭乱的磁头几何形状和定位机制^。52

IBM engineers, working at the newly established laboratory in San Jose, California, came up with the notion of using a thin film of air as a cushion to prevent the heads from touching the disk surface. Their first product used air supplied to the heads from an external compressor. Later IBM disk drives took advantage of an ingenious application of the “boundary layer”—an aerodynamic phenomenon familiar to airplane designers—to float the disk by air set in motion by the disk itself. The labs spent 1953 and 1954 experimenting with a bewildering variety of head geometries and positioning mechanisms.⁵²

到 1956 年，IBM 解决了这些问题，并公开发布了 305 型磁盘存储单元。它使用一叠 50 个铝盘，每个直径为 24 英寸，以 1200 rpm 的速度旋转。总存储容量为 500 万个字符。53^新闻稿强调了其“随机访问”的革命性品质：任何数据都可以像其他数据一样访问，这与从一副打孔卡或一卷磁带中顺序检索数据不同。该机器后来被重新命名为 RAMAC，即随机存取存储器会计机。IBM 还宣布，第一个商业客户，丹佛的联合航空公司，将使用它作为预订系统。1958 年春，IBM 在布鲁塞尔世界博览会的美国馆安装了一台 305，参观者可以通过键盘向“RAMAC 教授”提问，并以十种语言中的任何一种获得答案（包括国际语，一种像世界语一样的人工语言）。⁵⁴布鲁塞尔的展览是 IBM 公关的杰作。就像乔治·斯蒂比茨 1940 年在美国数学学会达特茅斯会议上进行的著名远程计算演示一样，它预示着有一天，直接访问大量数据将成为银行、超市、保险公司和政府机构运营的必需品。IBM 总裁托马斯·沃森 (Thomas Watson Jr.) 将 RAMAC 的推出称为“IBM 历史上最伟大的产品日”，但即使是他也没有完全理解 RAMAC 教授所引发的力量。随着时间的推移，随机存取磁盘内存实现的交互式计算方式将迫使 IBM 以及计算机行业的其他公司重新定义自己。

By 1956 IBM had solved the problems and publicly announced the Model 305 Disk Storage unit. It used a stack of fifty aluminum disks, each 24 inches in diameter, rotating at 1200 rpm. Total storage capacity was five million characters.⁵³ The press release emphasized its revolutionary quality of “random access”: that any piece of data was as accessible as any other, in contrast to the sequential retrieval of a datum from a deck of punched cards or a reel of tape. The machine was later rechristened RAMAC, for Random Access Memory Accounting Machine. IBM also announced that the first commercial customer, United Airlines in Denver, would use it for a reservations system. In the spring of 1958 IBM installed a 305 at the U.S. Pavilion at the Brussels World’s Fair, where visitors could query “Professor RAMAC” through a keyboard and receive answers in any of ten languages (including Interlingua, an artificial language like Esperanto).⁵⁴ The Brussels exhibit was a masterful piece of IBM public relations. Like the famous demonstration of remote computing staged by George Stibitz at the Dartmouth meeting of the American Mathematical Society in 1940; it foretold a day when direct access to large amounts of data would become essential to the operations of banks, supermarkets, insurance companies, and government agencies. IBM president, Thomas Watson Jr., heralded the RAMAC’s introduction as “the greatest product day in the history of IBM,” but even he did not fully understand the forces set in motion by Professor RAMAC. In time, the interactive style of computing made possible by random access disk memory would force IBM, as well as the rest of the computer industry, to redefine itself.

RAMAC 是作为 IBM 650 的独立设备提供的，与 650 一样，它也使用真空管。推出几年后，它就过时了，并从市场上撤出。但磁盘技术幸存下来，并进一步巩固了 IBM 在业界的主导地位。最终，允许用户直接与计算机交互的直接数据访问意味着批处理方法的终结，而 IBM 自制表机时代以来一直依靠这种方法建立自己的财务实力。

The RAMAC was offered as a separate piece of equipment to be attached to the IBM 650, and like the 650 it used vacuum tubes. Within a few years of its introduction it was obsolete and was withdrawn from the market. But the disk technology survived and furthered IBM’s dominance of the industry. Eventually direct access to data that allowed users to interact directly with a computer spelled the end of the batch method of processing, on which IBM had built its financial strength ever since the tabulator days.

从真空管到晶体管IBM 实力的第二个例子是它在开始交付 709 型号后仅一年就推出了 7090 型号。

From Vacuum Tubes to Transistors A second example of IBM’s strength is its introduction of model 7090 only a year after it began deliveries of model 709.

7090 是一款大型晶体管计算机，许多人认为它是经典的大型计算机，它表明 IBM 接受了真空管技术的过时，并愿意在 709 上承担经济损失。据传言，IBM 向美国空军提交了一份投标书，要求为北极圈周围的弹道导弹预警系统 (BMEWS) 提供固态计算机。当时 IBM 已经宣布了 709，但空军坚持使用晶体管机器，也许是因为其他较小的公司已经推出了受到行业媒体赞誉的晶体管产品。

The 7090, a large transistorized computer that many regard as the classic mainframe computer, showed that IBM accepted the obsolescence of vacuum tube technology and that it was willing to take a financial loss on the 709. According to folklore, IBM submitted a bid to the U.S. Air Force to supply solid state computers for the Ballistic Missile Early Warning System (BMEWS) around the Arctic Circle. At the time IBM had announced the 709, but the Air Force insisted on transistorized machines, perhaps because other, smaller companies had already introduced transistorized products that were getting praise from the trade press.

IBM 计划通过设计一台与 709 结构相同并使用晶体管的计算机来满足空军严格的时间表。因此，他们能够使用 709 来开发和测试新计算机所需的软件。709 被编程为“模拟”尚未完成的新机器：编写了一个程序使 709 表现得像新计算机一样。（此后，每当设计新计算机或计算机芯片时，这都成为标准程序。）即使这种技术也无法保证 IBM 能够在 1960 年之前满足空军的安装期限。IBM 于 1959 年底将计算机运送到格陵兰岛的一个站点，但 IBM 观察家声称这些机器尚未完成，该公司派遣了多达 200 名工程师前往格陵兰岛完成正在安装的机器^。55

IBM planned to meet the Air Force’s strict timetable by designing a computer that was architecturally identical to the 709 and used transistors. They were thus able to use a 709 to develop and test the software that the new computer would need. The 709 was programmed to “emulate” the as-yet-unfinished new machine: a program was written to make the 709 behave as if it were the new computer. (This has since become a standard procedure whenever a new computer or computer chip is being designed.) Even that technique did not guarantee that IBM would meet the Air Force’s deadline of installations before 1960. IBM delivered computers to a site in Greenland in late 1959, but IBM-watchers claimed that the machines were not finished and that the company dispatched a cadre of up to 200 engineers to Greenland to finish the machine as it was being installed.⁵⁵

无论这个故事是否属实，该公司确实向格陵兰交付了一台晶体管计算机，其版本以 7090 型的名称进行商业销售。7090 及其后来的升级版 7094（增加了四个索引寄存器）被视为经典的大型机，因为它兼具架构、性能和财务成功：安装了数百台机器，每台价格约为 200 万美元（图 2.5）。

Whether or not that story is true, the company did deliver a transistorized computer to Greenland, versions of which it marketed commercially as the Model 7090. The 7090 and its later upgrade, called the 7094, which had four additional index registers, is regarded as the classic mainframe because of its combination of architecture, performance, and financial success: hundreds of machines were installed at a price of around $2 million each (figure 2.5).

7094 安装描述“大型机” 一词可能源于这样一个事实：大型计算机的电路安装在大型金属框架上，并装在机柜中。框架装有铰链，可以旋转出来进行维护。典型的安装包括多个这样的机柜，这些机柜放在瓷砖地板上，比实际地板高出几英寸，留出空间用于连接从一个机柜到另一个机柜的众多粗连接电缆以及用于循环调节后的空气。整个房间可能都有自己的气候控制系统，该系统不仅与建筑物其他部分分开，而且在炎热的夏日里，使房间比建筑物其他任何地方都舒适得多。

A Description of a 7094 Installation The term “mainframe” probably comes from the fact that the circuits of a mainframe computer were mounted on large metal frames, housed in cabinets. The frames were on hinges and could swing out for maintenance. A typical installation consisted of a number of these cabinets standing on a tiled floor that was raised a few inches above the real floor, leaving room for the numerous thick connecting cables that snaked from one cabinet to another and for the circulation of conditioned air. The entire room probably had its own climate-control system that not only was separate from that of the rest of the building, but also kept the room much more comfortable than anywhere else in the building on a hot summer day.

靠近操作员控制台的柜子里安放着主处理器电路。这些电路由分立晶体管组成，与电阻器、二极管、跳线、电感器和电容器一起安装和焊接到印刷电路板上。这些电路板又被插入“背板”，其中复杂的线路网将信号从一个电路板传输到另一个电路板。有些主机是手工费力接线的，但大多数主机使用一种称为“绕接”的技术：它不需要焊接，对于生产机器来说，接线可以由机器完成，从而消除了错误。在实践中，总是会偶尔手工焊接跳线以纠正设计错误或以其他方式修改电路。这些电路的密度约为每立方英寸十个元件。

A cabinet near the operator’s console housed the main processor circuits. These were made up of discrete transistors, mounted and soldered along with resistors, diodes, jumper wires, inductors, and capacitors, onto printed circuit boards. The boards, in turn, were plugged into a “backplane,” where a complex web of wires carried signals from one circuit board to another. Some mainframes were laboriously wired by hand, but most used a technique called “wire wrap”: it required no soldering, and for production machines the wiring could be done by a machine, thus eliminating errors. In practice, there would always be occasional pieces of jumper wire soldered by hand to correct a design error or otherwise modify the circuits. The density of these circuits was about ten components per cubic inch.

图 2.5

IBM 7094 的控制台。请注意控制台顶部的小盒子，上面有四排指示灯，指示计算机的四个附加索引寄存器的状态。这些寄存器的添加是 7094 和 IBM 7090 之间的主要区别。还请注意其他几排指示灯，指示 CPU 其他寄存器中的位。底部的行标记为“存储”、“累加器”和“MQ”（表示乘商）。（来源：IBM。）

Figure 2.5

Control console of an IBM 7094. Note the small box on top of the console with four rows of lights, which indicate the status of the computer’s four additional index registers. The addition of these registers was the principal difference between the 7094 and the IBM 7090. Note also the other rows of lights indicating the bits in the other registers of the CPU. The rows at the bottom are labeled “Storage,” “Accumulator,” and “M-Q” (for multiply-quotient). (Source : IBM.)

7094 出厂时配备最大 32,768 个字的核心内存。按照现代术语来说，这相当于大约 150 千字节 (KB，1 字节 ¼ 8 位)，大约是 20 世纪 80 年代初 IBM 个人计算机首次亮相时的内存容量。虽然它被宣传为科学和工程用机器，但许多客户发现它非常适合执行各种任务。它每秒可以执行大约五十到十万次浮点运算，是当时最快的浮点运算之一。很难与现代计算机进行比较，因为标准已经改变，但它的速度大约与 20 世纪 80 年代后期的个人计算机一样快。它的 36 位字长适合需要许多位精度的科学计算，并且它还具有允许处理器直接寻址大量内存的优势。相比之下，第一台个人电脑仅使用 8 位字长；到 1990 年已成为标准的 32 位长度仍然比 7904 机器的长度短。

The 7094 was delivered with a maximum of 32,768 words of core memory. In modern terms that corresponds to about 150 Kilobytes (Kbytes, one byte ¼ 8 bits), about what came with the IBM Personal Computer when it first appeared in the early 1980s. Although marketed as a machine for science and engineering, many customers found it well suited for a variety of tasks. It could carry out about fifty to one-hundredthousand floating-point operations per second, making it among the fastest of its day. Comparisons with modern computers are difficult, as the yardsticks have changed, but it was about as fast as a personal computer of the late 1980s. Its 36-bit word length suited it for scientific calculations that require many digits of precision, and it had the further advantage of allowing the processor to address a lot of memory directly. By comparison, the first personal computers used only an 8-bit word length; the 32-bit length that had become a standard by 1990 was still shorter than what the 7904 machine had.

控制台本身装饰着一排令人印象深刻的闪烁指示灯、刻度盘、仪表和开关。它看起来就像人们所认为的计算机应该有的样子。甚至到了 20 世纪 90 年代，一些好莱坞电影也将计算机描绘成这样。（其他一些电影，如《侏罗纪公园》，则展示了普通的 UNIX 工作站。）一排排小指示灯指示组成计算机中央处理器的各个寄存器的每个位的状态。如果发生硬件故障或编程错误，操作员可以直接以二进制数形式读取每个寄存器的内容。他们还可以一步一步执行程序，并记录每一步寄存器的内容。如果需要，他们可以通过翻转开关直接更改寄存器的位。这种“位操作”非常繁琐，但它让操作员可以亲密地掌控机器，自那时起很少有人能享受这种乐趣。

The console itself was festooned with an impressive array of blinking lights, dials, gauges, and switches. It looked like what people thought a computer should look like. Even into the 1990s some Hollywood movies portrayed computers like this. (A few others, like Jurassic Park, showed modest UNIX workstations.) Rows of small lights indicated the status of each bit of the various registers that made up the computer’s central processor. In the event of a hardware malfunction or programming error, operators could read the contents of each register directly in binary numbers. They could also execute a program one step at a time, noting the contents of the registers at each step. If desired, they could directly alter the bits of a register by flipping switches. Such “bit twiddling” was exceedingly tedious, but it gave operators an intimate command over the machine that few since that time have enjoyed.

大多数时候，操作员不需要做这些事情。计算机的真正控制是由程序员完成的，很少有程序员被允许进入计算机室。程序员在打孔卡上开发他们的工作，这些卡由一台小型 IBM 1401 计算机读取并传输到一卷磁带上。操作员拿起这盘磁带并将其安装在连接到主机的磁带驱动器上（尽管有一个直接连接的读卡器供偶尔使用）。许多程序员很少看到实际运行程序的机器。事实上，许多程序员甚至不使用打字机，而是将他们的程序写在特殊的编码表上，然后交给打字机操作员。56^操作员的工作包括安装和拆卸磁带、不时按下按钮开始工作、偶尔将一叠卡片插入阅读器以及从打印机读取状态信息。这不是一个特别有趣或地位高的工作，尽管对于外行来说它看起来令人印象深刻。在 7904 的鼎盛时期，许多需要操作员判断的工作都被控制程序接管，这个控制程序被恰当地称为“操作系统”。

Most of the time, an operator had no need to do those kinds of things. The real controlling of the computer was done by its programmers, few of whom were ever allowed in the computer room. Programmers developed their work on decks of punched cards, which were read by a small IBM 1401 computer and transferred to a reel of tape. The operator took this tape and mounted it on a tape drive connected to the mainframe (although there was a card reader directly attached for occasional use). Many programmers seldom saw the machine that actually ran the programs. In fact, many programmers did not even use a keypunch, but rather wrote out their programs on special coding sheets, which they gave to keypunch operators.⁵⁶ The operator’s job consisted of mounting and demounting tapes, pressing a button to start a job every now and then, occasionally inserting decks of cards into a reader, and reading status information from a printer. It was not a particularly interesting or high-status job, though to the uninitiated it looked impressive. Over the course of the 7904’s heyday, many of the operator’s jobs that required judgment were taken over by a control program, appropriately called an “operating system.”

7094 的月租约为 30,000 美元，相当于 160 万美元的购买价。考虑到这一成本，这台机器绝对不能闲置。虽然我们的个人电脑会在我们开会或吃午饭时运行屏幕保护程序，但这种做法浪费的计算机周期数量在 1963 年是惊人的。在 7094 上，程序被收集到磁带卷上并分批运行。程序员必须等到一批程序运行完毕才能获得结果，如果他们发现自己犯了错误或需要进一步优化问题，他们必须提交新的程序并再次等待。无论多么诱人，直接访问机器（向其提交程序并等待其运行几秒钟）的想法都是不可能的，因为让处理器闲置几分钟的成本很高。这种操作方法是大型机时代的一个决定性特征。

A 7094 installation rented for about $30,000 a month, or an equivalent purchase price of about $1.6 million. With that cost it was imperative that the machine never be left idle. Although our personal computers run a screen-saver while we go to a meeting or to lunch, the number of computer cycles wasted by this practice would have been scandalous in 1963. On the 7094, programs were gathered onto reels of tape and run in batches. Programmers had to wait until a batch was run to get their results, and if they then found that they had made a mistake or needed to further refine the problem, they had to submit a new deck and wait once more. However tempting, the idea of gaining direct access to the machine—to submit a program to it and wait a few seconds while it ran—was out of the question, given the high costs of letting the processor sit idle for even a few minutes. That method of operation was a defining characteristic of the mainframe era.

磁带驱动器与处理器电路板一起主导着大型机的安装。这些磁带是将大型计算机与外界连接起来的媒介。程序和数据通过磁带输入计算机；作业的结果同样被发送到磁带。如果程序成功运行，操作员会拿起磁带并将其移动到连接到 1401 计算机的驱动器上，该驱动器就像个人计算机时代的“智能”打印机一样，处理链式打印机上打印结果的较慢过程。（与现代打印机不同，通常没有直接连接。）结果以大写字母打印在 15 英寸宽的扇形折叠纸上。

Along with the processor circuit cabinets, magnetic tape drives dominated a mainframe installation. These tapes were the medium that connected a mainframe computer to the outside world. Programs and data were fed into the computer through tapes; the results of a job were likewise sent to a tape. If a program ran successfully, an operator took the tape and moved it to the drive connected to a 1401 computer, which, like a “smart” printer of the Personal Computer era, handled the slower process of printing out results on a chain printer. (Unlike a modern printer, there was typically no direct connection.) Results were printed, in all capital letters, on 15-inch wide, fan-folded paper.

一些大型机配有视频控制台，但 7094 的主控制面板上没有。这样的控制台仅用于控制目的，因为磁带上的顺序存储无论如何都无法直接访问数据。一般来说，由于它们对核心内存的需求巨大，因此不使用它们。

A few mainframes had a video console, but there was none on the 7094’s main control panel. Such a console would have been useful only for control purposes, since the sequential storage on tapes prevented direct access to data anyway. In general, they were not used because of their voracious appetite for core memory.

随着个人计算机和工作站的出现，大型机通常被视为在世纪之交不会幸存的恐龙。事实上，大型机由于其存储和移动大量数据的能力而幸存下来，甚至蓬勃发展。但是，顺序、批量导向的操作模式，以及其特有的磁带库、穿孔卡片组和打印输出，已不再像 20 世纪 60 年代那样占据主导地位，当时没有其他选择。

With the advent of personal computers and workstations, mainframes are often viewed as dinosaurs that will not survive the turn of the century. In fact, the mainframe has survived and even prospered, because of its ability to store and move large quantities of data. But the sequential, batch-oriented mode of operation, with its characteristic libraries of tapes, decks of punched cards, and printouts, has ceased to dominate, as it did in the 1960s when there was no alternative.

小型晶体管机器

Small Transistorized Machines

20 世纪 50 年代末，晶体管技术的成熟也对计算机行业的低端市场产生了影响。固态计算机的成本低廉，性能却远高于几年前的鼓式计算机，这成为可能。晶体管的使用也使这些机器体积更小，可以安装在以前放置制表机和打卡机的同一间房间内，而这些机器通常用于处理公司的数据处理需求。

The maturing of transistor technology by the end of the 1950s also had an impact at the low end of the computer industry. It became possible to offer solid state computers at low cost and with much better performance than the drum machines of a only a few years earlier. The use of transistors also made these machines compact, allowing them to be installed in the same rooms that previously housed the array of tabulators and card punches that typically handled a company’s data-processing needs.

同样，这些机器中最成功的是 IBM 于 1959 年推出的 1401 型（图 2.6）。如果说 650 证明了计算机市场潜力巨大，那么面向商业客户的 1401 则表明市场确实很大，几乎涵盖了所有使用打孔卡设备的地方。最终安装了超过一万台 1401 — 相比之下，UNIVAC 和 IBM 701 仅有 30 到 40 台，而 IBM 650 的安装量约为一千台。与此同时，IBM 还为科学客户推出了 1620，这是一款架构略有不同的小型计算机，同样销量不错。

Once again, the most successful of these machines came from IBM, the model 1401, announced in 1959 (figure 2.6). If the 650 had demonstrated that the market for computers was potentially very large, the 1401, intended for business customers, showed that the market was indeed a real one and included nearly every place where punched card equipment was used. Eventually over ten thousand 1401s were installed—compare this number to the thirty to forty UNIVACs and IBM 701s, or the approximately one thousand installations of the IBM 650. At the same time IBM also introduced the 1620, a small computer with a slightly different architecture, for scientific customers, and it, too, sold well.

1401 在早期开发阶段一直采用插接板架构。与 650 的设计一样，1955-1956 年，IBM 工程师对其进行了重新设计，使其融入了存储程序架构，从而可以将其编程为通用计算机。57 ¹⁶²⁰将这一设计发挥到了极致：其大部分算术运算不是通过连接到处理器的逻辑电路完成的，而是通过引用存储在更便宜（但更慢）的核心内存中的算术表完成的。一些精明的客户甚至通过简单地将不同的数字存储在计算机查找数字总和的内存位置中来更改 1620 的指令集和逻辑。他们说非正式名称“CADET”是“无法添加；甚至不尝试！”的首字母缩写词，但它确实有效。（17 世纪 20 年代利用这样一组原始电路进行算术运算的能力后来启发了微处理器的一位发明者在一片硅片上进行同样的运算。）

The 1401 had a plugboard architecture throughout its early phases of development. As happened with the 650’s design, in 1955–1956 IBM engineers redesigned it to incorporate a stored-program architecture which allowed it to be programmed as a general-purpose computer.⁵⁷ The 1620 carried this to an extreme: most of its arithmetic was done not by logic circuits wired into the processor, but by references to arithmetic tables stored in the cheaper (but slower) core memory. Some savvy customers even altered the 1620’s instruction set and logic by the simple act of storing different numbers in the memory locations the computer looked to for the sums of numbers. They said that the informal name “CADET” was an acronym for “Can’t Add; Doesn’t Even Try!” But it worked. (The 1620’s ability to do arithmetic with such a primitive set of circuits would later inspire one of the inventors of the microprocessor to do the same on a sliver of silicon.)

1401 的性能一般。由于字长可变，其处理速度也各不相同，平均而言，它只比 650 快七倍左右。与 650 不同，它很少用于科学或工程应用，尽管大多数科学主机安装都使用一台或多台 1401 将数据从卡片传输到磁带并打印。

The 1401 offered modest performance. With its variable word length, its processing speed varied, and on average it was only about seven times faster than a 650. Unlike the 650 it was rarely used for scientific or engineering applications, although most scientific mainframe installations used one or more to transfer data from cards to tape and to print.

图 2.6

IBM 1401 生产线。1401 是第一台产量可与 IBM 穿孔卡会计机生产线相媲美的电子数字计算机。（来源：IBM。）

Figure 2.6

IBM 1401 production line. The 1401 was the first electronic digital computer to be produced in quantities comparable to IBM’s line of punched-card accounting machines. (Source : IBM.)

无论是单独使用还是作为大型计算机的附件，1401 的成功很大程度上要归功于 IBM 随其推出的一款外围设备，即 1403 型打印机。这款打印机使用在页面上横向移动的连续字符链。磁力驱动的锤子将链条敲击在需要打印字符的精确位置。它每分钟能够打印 600 行，比市场上任何其他产品都要快得多，而且坚固耐用，可以承受频繁使用。58^这款打印机使用老式机械技术来补充主机中进行的奇特电子处理。它是一种实用设备，但用户对它有着非理性的喜爱。几乎在每一所大学的计算机中心，都有人想出了如何通过向打印机发送适当的命令来对打印机进行编程，使其播放学校的战歌。声音质量很差，但打印机并没有被要求播放勃拉姆斯。其他人可能会用它来打印史努比的粗糙图像作为一系列字母字符。在好莱坞内外，喋喋不休的链式打印机、旋转的磁带和闪烁的灯光成为了计算机时代的象征。

Whether used alone or as an adjunct to a large computer, the 1401’s success owed a lot to a piece of peripheral equipment that IBM introduced with it, the type 1403 printer. This printer used a continuous chain of characters that moved laterally across the page. Magnetically driven hammers struck the chain at the precise place where a desired character was to be printed. Capable of printing 600 lines per minute, it was much faster than anything else on the market, and it was rugged enough to stand up under heavy use.⁵⁸ The printer used old-fashioned mechanical technology to complement the exotic electronic processing going on in the mainframe. It was a utilitarian device but one that users had an irrational affection for. At nearly every university computer center, someone figured out how to program the printer to play the school’s fight song by sending appropriate commands to the printer. The quality of the sound was terrible, but the printer was not asked to play Brahms. Someone else might use it to print a crude image of Snoopy as a series of alphabetic characters. In and out of Hollywood, the chattering chain printer, spinning tapes, and flashing lights became symbols of the computer age.

结论

Conclusion

到 1960 年，商业计算的模式已经确立，这种模式将持续到接下来的二十年。需求最大的客户在特殊的气候控制室中安装了大型主机，由一群技术人员负责管理。这些主机使用核心内存，并辅以磁盘或磁鼓组。支持这些内存的是磁带驱动器组，以及一个用于存档磁带卷的图书馆。尽管磁盘和磁鼓允许随机访问数据，但大多数访问都符合磁带和卡片组上数据存储的顺序性。

By 1960 a pattern of commercial computing had established itself, a pattern that would persist through the next two decades. Customers with the largest needs installed large mainframes in special climate-controlled rooms, presided over by a priesthood of technicians. These mainframes utilized core memories, augmented by sets of disks or drums. Backing that up were banks of magnetic tape drives, as well as a library where reels of magnetic tape were archived. Although disks and drums allowed random access to data, most access conformed to the sequential nature of data storage on tapes and decks of cards.

对于大学环境中的大多数用户来说，典型的交易是从通过窗口向操作员提交一副纸牌开始的（以保持计算机室的气候控制）。一段时间后，用户前往打印机输出的地方，取回包含其作业结果的折叠纸块。打印输出的前几页用于解释作业花费了多长时间、使用了多少内存、访问了哪个磁盘或磁带驱动器等 - 这些信息对计算机中心的操作员很有用，而且写得非常神秘，足以吓倒任何未入教的用户。

For most users in a university environment, a typical transaction began by submitting a deck of cards to an operator through a window (to preserve the climate control of the computer room). Sometime later the user went to a place where printer output was delivered and retrieved the chunk of fan-fold paper that contained the results of his or her job. The first few pages of the printout were devoted to explaining how long the job took, how much memory it used, which disk or tape drives it accessed, and so on—information useful to the computer center’s operators, and written cryptically enough to intimidate any user not initiated into the priesthood.

对于商业和工业计算机中心，此过程更为常规，但本质上是相同的。计算机中心通常会定期（例如，每周一次）运行一组程序，使用键盘操作员提供的新数据。处理这些数据的程序可能会在每次运行之间略有变化，尽管人们认为这是例外，而不是常态。打印件被“撕开”（沿穿孔撕开），装订在软封面之间，并放在滚动架或架子上。这些打印件为组织提供了做出决策和开展日常运营所需的数据。

For commercial and industrial computer centers, this procedure was more routine but essentially the same. The computer center would typically run a set of programs on a regular basis—say, once a week— with new data supplied by keypunch operators. The programs that operated on these data might change slightly from one run to the next, although it was assumed that this was the exception rather than the rule. The printouts were “burst” (torn along their perforations), bound between soft covers, and placed on rolling racks or on shelves. These printouts supplied the organization with the data it needed to make decisions and to run its day to day operations.

因此，早期的计算时代以批处理为特征。硬件的成本使得用户无法像今天这样与计算机交互。直接交互访问计算机数据并非闻所未闻，但仅限于成本不是影响因素的应用程序，例如 SAGE 防空系统。对于商业客户来说，批处理并不是一个严重的障碍。依赖几天后才更新的打印报告与社会其他领域的运输和通信速度并不相符。批处理的缺点，尤其是它使编写和调试程序变得困难，在教授计算机编程学科的大学中更为明显。因此，大学教师和学生认识到需要将交互式计算带入主流。在接下来的几年里，这一需求将得到满足，尽管这将是一个漫长而艰难的过程。

Thus the early era of computing was characterized by batch processing. The cost of the hardware made it impractical for users to interact with computers as is done today. Direct interactive access to a computer’s data was not unknown but was confined to applications where cost was not a factor, such as the SAGE air defense system. For business customers, batch processing was not a serious hindrance. Reliance on printed reports that were a few days out of date was not out of line with the speeds of transportation and communication found elsewhere in society. The drawbacks of batch processing, especially how it made writing and debugging programs difficult, were more noticed in the universities, where the discipline of computer programming was being taught. University faculty and students thus recognized a need to bring interactive computing to the mainstream. In the following years that need would be met, although it would be a long and difficult process.

表 2.2列出了本章讨论的一些机器的特性。

Table 2.2 lists the characteristics of some of the machines discussed in this chapter.

表 2.2

本章讨论的选定计算机的特征

Table 2.2

Characteristics of selected computers discussed in this chapter

姓名	发布或安装年份	字长	主记忆单词	设备类型
智者	1955–1958	32 位	八千	管
飞歌TRANSAC-2000	1958	—	—	晶体管
RCA 501	1958	12 位小数		晶体管
IBM 1401	1959	变量（7位/字符）	4–16 千	晶体管
IBM 7090	1960	36 位	三十二 K	晶体管

软件的早期历史，1952-1968

The Early History of Software, 1952–1968

他拥有镇上最好的商店，并且做着一笔不错的软件生意，尤其是在圣诞节期间驮马安全归来的时候。

He owned the very best shop in town, and did a fine trade in soft ware, especially when the pack horses came safely in at Christmas-time.

—RD Blackmore、Lorna Doone ¹

—R. D. Blackmore, Lorna Doone¹

这个人的军队里不会有软件！

There will be no software in this man’s army!

— 德怀特·艾森豪威尔将军，约 1947 年²

—General Dwight D. Eisenhower, ca. 1947²

1993 年，美国工程院将查尔斯·斯塔克·德雷珀奖授予约翰·巴克斯，以表彰他“开发了 FORTRAN ……第一个通用高级计算机语言”。3^大多数计算机专业人士都对工程院的选择表示赞赏，他们中的大多数人都深知巴克斯和 FORTRAN 对他们行业的贡献。FORTRAN 虽然目前仍在使用，但早已被许多其他语言（如 C++ 或 Visual Basic）以及 UNIX 和 Windows 等系统软件所取代，这些软件反映了个人计算机和工作站不断变化的硬件环境。在接受奖项时，巴克斯慷慨地承认这是团队努力的结果，并列举了几位同事为 FORTRAN 的诞生付出了长期艰苦的努力。

In 1993 the National Academy of Engineering awarded its Charles Stark Draper Prize to John Backus, “for his development of FORTRAN … the first general-purpose, high-level computer language.”³ The Academy’s choice was applauded by most computer professionals, most of whom knew well the contribution of Backus and FORTRAN to their profession. FORTRAN, although currently still in use, has long been superseded by a host of other languages, like C+ + or Visual Basic, as well as by system software such as UNIX and Windows, that reflect the changing hardware environment of personal computers and workstations. In accepting the award, Backus graciously acknowledged that it was a team effort, and he cited several coworkers who had labored long and hard to bring FORTRAN into existence.

设立德雷珀奖是为了给予工程师们与诺贝尔奖给予科学家一样的声望和奖金。在这里，该奖项颁发给开发软件的人——软件从定义上来说没有物理实体，恰恰不是“硬件”。当电子计算机刚刚开发出来时，很少有人认为该奖项是必需的。不仅事实证明 FORTRAN 之类的软件是必需的，而且到 20 世纪 90 年代，它的开发和营销甚至超过了硬件，在某些情况下，硬件正在成为一种廉价的批量生产商品。现在被称为“软件”的实体是如何出现的？它与计算机硬件的发展有何关系？

The Draper Prize was instituted to give engineers the prestige and money that the Nobel Prize gives scientists. Here it was being awarded for developing a piece of software—something that, by definition, has no physical essence, precisely that which is not “hardware.” The prize was being awarded for something that, when the electronic computer was first developed, few thought would be necessary. Not only did it turn out that software like FORTRAN was necessary; by the 1990s its development and marketing overshadowed hardware, which was becoming in some cases a cheap mass-produced commodity. How did the entity now called “software” emerge, and what has been its relationship to the evolution of computer hardware?

软件的简单定义是，它是指导计算机执行特定任务的一组指令。每台机器都有它。在十九世纪的运河中，拖船通过船闸需要执行一系列精确的步骤，每个步骤都必须以正确的顺序和正确的方式完成。对于运河船来说，有两套程序：一套是将船从低处移到高处，另一套是将船从高处移到低处。这些步骤可以正式化并写下来，但没有运河工人称它们为“软件”。这并不是因为这些程序很简单，而是因为它们与船闸的唯一目的密切相关：将运河船从一个水位段移到另一个水位段。运河船闸可能有次要用途，比如提供灌溉用水，但这些都不是设计或安装船闸的原因。

A simple definition of software is that it is the set of instructions that direct a computer to do a specific task. Every machine has it. Towing a boat through a lock of a nineteenth-century canal required a performing sequence of precise steps, each of which had to be done in the right order and in the right way. For canal boats there were two sets of procedures: one for getting a boat from a lower to a higher level, and one for going the other way. These steps could be formalized and written down, but no canal workers ever called them “software.” That was not because the procedures were simple, but because they were intimately associated with the single purpose of the lock: to get a canal boat from one level stretch to another. A canal lock may have secondary purposes, like providing water for irrigation, but these are not the reasons the lock is designed or installed.

相比之下，计算机没有指定要解决的任何单个问题。没有主要功能和次要功能的划分：存储程序数字计算机本质上是一种通用机器，这就是为什么用户的程序更为重要。这些程序应被视为与运行它们的机器分开。

A computer, by contrast, does not specify any single problem to be solved. There is no division into primary and secondary functions: a stored-program digital computer is by nature a general-purpose machine, which is why the procedures of users assume greater importance. These procedures should be considered separate from the machine on which they run.

“软件”一词表明存在一个独立于计算机硬件的单一实体，它与硬件协同解决问题。事实上，并不存在这样的单一实体。计算机系统就像一个洋葱，在硬件核心上有许多不同的软件层。即使在中心——中央处理器层面——也没有明显的区别：带有“微代码”的计算机芯片指示其他芯片执行处理器最基本的操作。工程师将这些代码称为“固件”，这个术语表明了这种模糊的区别。

The word “software” suggests that there is a single entity, separate from the computer’s hardware, that works with the hardware to solve a problem. In fact, there is no such single entity. A computer system is like an onion, with many distinct layers of software over a hardware core. Even at the center—the level of the central processor—there is no clear distinction: computer chips carrying “microcode” direct other chips to perform the processor’s most basic operations. Engineers call these codes “firmware,” a term that suggests the blurred distinction.

如果微代码在一端，那么在另一端，就会遇到类似自动柜员机 (ATM) 的东西，客户按下一系列按钮，使复杂的计算机网络正确地执行一组复杂的操作。ATM 的设计者认为用户对计算机知之甚少，但同样，客户正在对银行的计算机进行编程。使用 ATM 具有更广泛意义上的编程的许多属性。例如，在一长串按键中，只要按错一个键，就可能导致整个交易无效，而设计不良的 ATM 甚至会让计算机知识渊博的客户感到困惑（例如家用录像机，大多数用户发现无法对其进行编程）。

If microcode is at one end, at the other one encounters something like an automatic teller machine (ATM), on which a customer presses a sequence of buttons that causes a sophisticated computer network to perform a complex set of operations correctly. The designers of ATMs assume that users know little about computers, but just the same, the customer is programming the bank’s computer. Using an ATM shares many of the attributes of programming in the more general sense. Pressing only one wrong key out of a long sequence, for example, may invalidate the entire transaction, and a poorly designed ATM will confuse even a computer-literate customer (like the home video-cassette recorder, which most owners find impossible to program).

软件的本质就存在于这两个极端之间。程序员斯科特·金 (Scott Kim) 曾说：“编程计算机和使用计算机之间没有根本区别。” ⁴对他来说，这些层次是平滑连续的，从固件中嵌入的微代码到 ATM 的菜单命令，而他自己的工作则介于两者之间。（金是个人计算机软件设计师。）其他人则不那么肯定。开发复杂系统软件的人经常说，他们的工作与学校教授的计算机编程几乎没有关系。更糟糕的是，他们觉得教授计算机编程的方式是使用简单的例子，这让学生误以为软件的制作比实际要简单得多。5^他们还指出，开发好的软件并不只是编写好的单个程序，而是编写各种在复杂系统中彼此良好交互的程序。

Somewhere between these extremes lies the essence of software. One programmer, Scott Kim, said that “there is no fundamental difference between programming a computer and using a computer.”⁴ For him the layers are smooth and continuous, from the microcode embedded in firmware to the menu commands of an ATM, with his own work lying somewhere in the middle. (Kim is a designer of personal computer software.) Others are not so sure. People who develop complex system software often say that their work has little to do with the kind of computer programing taught in schools. What is worse, they feel that the way computer programming is taught, using simple examples, gives students a false sense that the production of software is a lot simpler than it is.⁵ They also point out that developing good software is not so much a matter of writing good individual programs as it is of writing a variety of programs that interact well with each other in a complex system.

软件史不应与计算史分开对待，即使这种区分对计算机工程师或科学家来说很有价值（3.1）。以下几个例子将展示软件创新在与硬件的相应创新很好地融合之前几乎没有影响。6^同样，人们经常观察到的现象是，硬件的进步（以硅片上的电路数量等指标衡量）远远超过软件的进步，这可能是错误的。7^虽然硬件技术必须面对和克服物理、有形的限制，但两者都必须面对和克服设计复杂性这一更具限制性的障碍^。8

The history of software should not be treated separately from the history of computing, even if such a distinction is of value to computer engineers or scientists ( 3.1). Several of the examples that follow will show innovations in software that had little or no impact until they could mesh well with corresponding innovations in hardware.⁶ Likewise, the often-repeated observation that progress in hardware, measured by metrics like the number of circuits on a silicon chip, far outpaces progress in software is probably false.⁷ While it is true that hardware technology must face and overcome limits of a physical, tangible nature, both face and overcome the much more limiting barrier of complexity of design.⁸

开始阶段（1944–1951）

Beginnings (1944–1951)

为了给哈佛 Mark I 机电计算机编程，用户在纸带上打了一排孔（每行最多 24 个），用于执行每条指令。9 ¹⁹⁴⁴年夏天，当这台机器公开亮相时，海军命令 Grace Murray Hopper 到计算实验室协助 Howard Aiken 编程。Hopper 曾是瓦萨学院的数学教授，后来请假去海军海军中校学习。据 Hopper 说，她刚到哈佛实验室报到时，刚刚获得一条纹半。在那里，Howard Aiken 向她展示了一个带有三条纹的大物体……挥了挥手说：“那是一台计算机。”我说：“是的，长官。”我还能说什么呢？他说他想让我计算反正切级数的系数，星期四。同样，我能说什么呢？“是的，长官。”我不知道到底发生了什么，但这就是我与 Howard Hathaway Aiken 的会面。¹⁰

In order to program the electromechanical Harvard Mark I, users punched a row of holes (up to 24 in a line) on a piece of paper tape for each instruction.⁹ In the summer of 1944, when the machine was publicly unveiled, the Navy ordered Grace Murray Hopper to the Computation Lab to assist Howard Aiken with programming it. Hopper had been a professor of mathematics at Vassar College and had taken leave to attend the Navy’s Midshipmen School. According to Hopper, she had just earned her one and one-half stripes when she reported to the lab at Harvard. There, Howard Aiken showed her a large object, with three stripes … waved his hand and said: “That’s a computing machine.” I said, “Yes, Sir.” What else could I say? He said he would like to have me compute the coefficients of the arc tangent series, for Thursday. Again, what could I say? “Yes, Sir.” I did not know what on earth was happening, but that was my meeting with Howard Hathaway Aiken.¹⁰

图 3.1

典型系统的软件与硬件的相对成本，1965-1985 年。这个著名的图表在 20 世纪 70 年代初由当时在 TRW 工作的 Barry Boehm 推广。该图表被转载到许多有关软件开发的教科书和文章中，并成为软件的一大神话。与任何神话一样，该图表也有很多真实性，但最近对软件支出的研究似乎得出这样的结论：多年来，该比率基本保持不变。（来源：改编自 Barry Boehm 的《软件及其影响》， Datamation [1973 年 5 月]：49。）

Figure 3.1

Relative costs of software vs. hardware for typical systems, 1965–1985. This famous graph was popularized in the early 1970s by Barry Boehm, then of TRW. The graph has been reprinted in numerous textbooks and articles about software development and has become one of the great myths of software. As with any myth there is much truth in this graph, but more recent studies of software expenditures seem to conclude that over the years the ratio has remained more or less constant. (Source : Adapted from Barry Boehm, “Software and its Impact,” Datamation [May 1973]: 49.)

从此，计算机编程在美国开始流行。霍珀为该系列编写了代码序列，后来又为更复杂的数学表达式编写了代码——最早的数学表达式之一是詹姆斯·贝克（哈佛研究员，因设计美国情报机构使用的绝密相机镜头而在圈内广为人知）的镜头设计问题^。11

Thus began the practice of computer programming in the United States. Hopper wrote out the sequence of codes for that series, and later the codes for more complex mathematical expressions—one of the first was a lens design problem for James Baker (a Harvard Fellow known among insider circles for his design of lenses for the top-secret cameras used by U.S. intelligence agencies).¹¹

一些反复使用的序列被永久地连接到 Mark I 的电路中。但这些序列很少，而且它们的使用并没有显著地扩展它的灵活性。由于 Mark I 不是存储程序计算机，霍珀别无选择，只能在连续的磁带上编码相同的模式。12^她很快就意识到，如果能找到一种方法来重复使用已经为另一个问题编码的磁带，那么就会节省很多精力。Mark I 并不容易做到这一点，但这个想法已经扎根，后来的修改确实允许安装多个磁带循环。

Some sequences that were used again and again were permanently wired into the Mark I’s circuits. But these were few and their use did not appreciably extend its flexibility. Since the Mark I was not a stored-program computer, Hopper had no choice for other sequences than to code the same pattern in successive pieces of tape.¹² It did not take long for her to realize that if a way could be found to reuse the pieces of tape already coded for another problem, a lot of effort would be saved. The Mark I did not allow that to be easily done, but the idea had taken root and later modifications did permit multiple tape loops to be mounted.

在哈佛大学后来的计算器（Mark III）的设计中，霍华德·艾肯开发了一种装置，它接收程序员在键盘上以普通数学符号输入的命令，然后将其转换成 Mark III 可以执行的数字代码（图 3.2）。这些代码记录在磁带上，然后输入 Mark III 并执行。常用的序列存储在磁鼓上。在德国，康拉德·楚泽也独立提出了类似的想法：他设想了一种“计划准备机”（Planfertigungsgeräte），可以为他在二战期间制造的 Z4 计算机打孔。13^楚泽的设备不仅可以翻译命令，还可以检查用户的输入以确保其语法正确，即左右括号的数量是否相同，两个数字之间是否没有出现多个算术运算，等等。

In the design of a later Harvard calculator (the Mark III), Howard Aiken developed a device that took a programmer’s commands, typed on a keyboard in the notation of ordinary mathematics, and translated them into the numerical codes that the Mark III could execute (figure 3.2). These codes, recorded on a magnetic tape, were then fed into the Mark III and carried out. Frequently used sequences were stored on a magnetic drum. In Germany, Konrad Zuse had independently proposed a similar idea: he had envisioned a “Plan Preparation Machine” (Planfertigungsgeräte) that would punch tapes for his Z4 computer, built during World War II.¹³ Zuse’s device would not only translate commands but also check the user’s input to ensure that its syntax was correct, that is, that it had the same number of left and right parentheses, that more than one arithmetic operation did not appear between two numbers, and so on.

图 3.2

哈佛 Mark III 上的编程机，约 1952 年。操作员是苏黎世联邦技术学院的 Ambros P. Speiser 教授。程序将以类似于普通代数的语言输入到此设备中，然后机器会将其翻译成 Mark III 可以执行的代码。使用存储程序计算机，这块额外的硬件是不必要的。（来源：德国波恩数学与数据处理协会 [GMD]。）

Figure 3.2

A programming machine attached to the Harvard Mark III, ca. 1952. The operator is Professor Ambros P. Speiser, of the Federal Technical Institute of Zurich. Programs would be keyed into this device in a language similar to ordinary algebra, and the machine would translate it into the codes that the Mark III proper could execute. With a stored program computer this additional piece of hardware is unnecessary. (Source : Gesellschaft für Mathematik und Datenverarbeitung [GMD], Bonn, Germany.)

祖泽从未完成计划准备机，尽管他在 1952 年改进了它的设计（并将其改名为“程序员”）。与此同时，他的 Z4 计算机已翻新并安装在苏黎世联邦技术学院。在那里使用它时，海因茨·鲁蒂豪瑟意识到一个重要事实：通用计算机本身可以被编程为像这样的“程序员”一样运行，从而无需单独的机器。因此，解决问题需要两个步骤：第一步是编程计算机以检查和翻译用户的命令，另一步是执行这些命令，这些命令现在以数字代码编码在磁带上。14^鲁蒂豪瑟简单地说：“将计算机用作自己的计划准备机。” ¹⁵

Zuse never completed the Plan Preparation Machine, although he had refined its design (and changed its name to “Programmator”) by 1952. Meanwhile, his Z4 computer had been refurbished and installed at the Federal Technical Institute in Zurich. While using it there, Heinz Rutishauser recognized an important fact: that a general-purpose computer could itself be programmed to act like such a “Programmator,” getting rid of the need for a separate machine. Solving a problem would thus take two steps: one in which the computer is programmed to check and translate the user’s commands, and another to carry out these commands, which are now encoded in numerical code on a piece of tape.¹⁴ Rutishauser stated it simply: “Use the computer as its own Plan Preparation Machine.”¹⁵

上述所有机器都没有将程序存储在内部存储器中，这意味着，像鲁蒂豪瑟设想的那样，对它们进行编程以翻译用户的命令将非常困难。然而，祖斯机器的设计灵活而优雅，这启发了鲁蒂豪瑟，让他清楚地看到了如何让计算机更容易编程。就像霍珀意识到她准备的磁带可以多次使用一样，鲁蒂豪瑟意识到解决问题的同一台计算机可以准备自己的指令，这是软件诞生的关键时刻。

None of the machines described above stored their programs in internal memory, which meant that programming them to translate a user’s commands as Rutishauser envisioned would have been very difficult. The Zuse machine, however, had a flexible and elegant design, which inspired Rutishauser to see clearly how to make computers easier to program. Like Hopper’s realization that the tapes she was preparing could be used more than once, Rutishauser’s realization that the same computer that solved a problem could prepare its own instructions was a critical moment in the birth of software.

使用存储程序计算机，可以将需要多次执行的指令序列存储在磁带上。当特定问题需要该序列时，计算机可以读取该磁带，将序列存储在内存中，并将序列插入程序中的适当位置。通过建立一个涵盖计算机最常用操作的序列库，程序员可以编写一个复杂的程序，而无需不断求助于指导机器的二进制代码。在早期的存储程序计算机中，英国剑桥的 EDSAC 将这一方案发挥到了极致，它有一个已经编写、开发和测试过的序列库，并打在纸带上，用户可以收集并合并到自己的程序中。16 ^EDSAC团队的 DJ Wheeler 设计了一种存储主程序的（不同）地址的方法，这些序列每次执行时都必须跳转到这些地址。这种所谓的 Wheeler Jump 是现代子程序调用的前身^。17

With a stored-program computer, a sequence of instructions that would be needed more than once could be stored on a tape. When a particular problem required that sequence, the computer could read that tape, store the sequence in memory, and insert the sequence into the proper place(s) in the program. By building up a library of sequences covering the most frequently used operations of a computer, a programmer could write a sophisticated and complex program without constant recourse to the binary codes that directed the machine. Of the early stored-program computers, the EDSAC in Cambridge, England, carried this scheme to the farthest extent, with a library of sequences already written, developed, and tested, and punched onto paper tapes that a user could gather and incorporate into his own program.¹⁶ D. J. Wheeler of the EDSAC team devised a way of storing the (different) addresses of the main program that these sequences would have to jump to and from each time they were executed. This so-called Wheeler Jump was the predecessor of the modern subroutine call.¹⁷

UNIVAC 编译器（1952）

UNIVAC Compilers (1952)

如果将这些序列打到卡片上，则可以通过选择适当的卡片、在卡片上写入和打出过渡代码，并将结果分组到新的卡片上来编写程序。这导致了“编译”一词的出现。到 20 世纪 50 年代初，计算机用户开发了允许计算机接管这些杂务的程序，这些程序被称为“编译器”。格雷斯·霍珀 (1906-1992) 在将这一概念从哈佛大学霍华德·艾肯实验室转移到商业世界方面发挥了至关重要的作用。尽管她希望继续留在军中，海军也为她提供了继续从事计算工作的机会，但约翰·莫奇利还是说服她加入他，并在 UNIVAC 建造时对其进行编程。18 ^（她最终回到海军现役，并在退休时晋升为海军少将。）¹⁹

If these sequences were punched onto decks of cards, a program could be prepared by selecting the appropriate decks, writing and punching transitional codes onto cards, and grouping the result on a new deck of cards. That led to the term “to compile” for such activity. By the early 1950s, computer users developed programs that allowed the computer to take over these chores, and these programs were called “compilers.” Grace Hopper (1906–1992) played a crucial role in transferring that concept from Howard Aiken’s laboratory at Harvard to the commercial world. Even though she had a desire to remain in uniform and the Navy had offered her continued employment in computing, John Mauchly was able to persuade her to join him and work on programming the UNIVAC as it was being built.¹⁸ (She eventually returned to active duty in the Navy and reached the rank of rear admiral at her retirement.)¹⁹

Hopper 将“编译器”定义为“一种程序编写例程，它可以为特定问题生成特定程序。” ²⁰她将使用编译器的整个活动称为“自动编程”。从 1952 年开始，一台名为“A-0”的编译器在 UNIVAC 上运行；1953 年，又推出了“A-1”和“A-2”。同年年底，UNIVAC 的客户可以使用 A-2 版本；据 Hopper 说，他们在几个月内就开始使用它了^。21

Hopper defined “compiler” as “a program-making routine, which produces a specific program for a particular problem.”²⁰ She called the whole activity of using compilers “Automatic Programming.” Beginning in 1952, a compiler named “A-0” was in operation on a UNIVAC; it was followed in 1953 by “A-1” and “A-2.” A version of A-2 was made available to UNIVAC’s customers by the end of that year; according to Hopper they were using it within a few months.²¹

如今，“编译器”一词已逐渐被广泛使用，指的是将用人类熟悉的语言编写的指令转换成计算机可以执行的二进制代码的程序。但霍珀的本意并不是这样。22 对^她来说，编译器处理存储在库中的子程序。23^根据霍珀的定义，编译方法是一种程序，它将子程序代码复制到主程序中程序员想要使用的适当位置。这些子程序的作用域有限，通常仅限于计算正弦、余弦、对数，尤其是浮点运算。尽管如此，编译器仍然是复杂的软件。要复制一个计算某个数字的对数的例程，需要指定要计算对数的数字的位置以及放置结果的位置，而结果通常每次程序使用这个特定的子程序时都会有所不同。²⁴虽然用子程序构建块“组装”程序的比喻很有说服力，但考虑到将子程序集成到无缝指令流中的难度，这种比喻并不恰当。自动编程支持者的目标是为软件开发亨利·福特为汽车生产开发的系统，即基于可互换零件的系统。但就像福特的系统在只生产一种车型时效果最好一样，这些早期的系统同样缺乏灵活性，它们试图过早地在错误的抽象层次上实现标准化。但只有在尝试的过程中，他们才意识到这一事实^。25

The term “compiler” has come into common use today to mean a program that translates instructions written in a language that human beings are comfortable with, into binary codes that a computer can execute. That meaning is not what Hopper had in mind.²² For her, a compiler handled subroutines stored in libraries.²³ A compiler method, according to Hopper’s definition, was a program that copied the subroutine code into the proper place in the main program where a programmer wanted to use it. These subroutines were of limited scope, and typically restricted to computing sines, cosines, logs, and, above all, floating-point arithmetic. Compilers nonetheless were complex pieces of software. To copy a routine that computed, say, the log of a number, required specifying the location of the number it was to calculate the log of, and where to put the results, which would typically be different each time a program used this specific subroutine.²⁴ The metaphor of “assembling” a program out of building blocks of subroutines, though compelling, was inappropriate, given the difficulty of integrating subroutines into a seamless flow of instructions. The goal for proponents of Automatic Programming was to develop for software what Henry Ford had developed for automobile production, a system based on interchangeable parts. But just as Ford’s system worked best when it was set up to produce only one model of car, these early systems were likewise inflexible, they attempted to standardize prematurely and at the wrong level of abstraction. But it was only in making the attempt that they realized that fact.²⁵

Laning 和 Zierler (1954)

Laning and Zierler (1954)

第一个以现代编译器的名义运行的编程系统是由 JH Laning 和 N. Zierler 于 20 世纪 50 年代初为麻省理工学院的 Whirlwind 计算机开发的。他们在一本优雅而简洁的手册中描述了这个从未命名的系统，该手册名为“Whirlwind I 的数学方程式翻译程序”，1954 年 1 月，麻省理工学院向大约一百个地点分发了这套系统。26^用John Backus 的话来说，它是“一个优雅的概念，优雅地实现了它”。与 UNIVAC 编译器不同，该系统的工作原理与现代编译器非常相似；也就是说，它将用户输入的命令作为输入，并生成新的机器代码作为输出，这些机器代码不仅执行这些命令，还跟踪存储位置、处理重复循环以及执行其他日常工作。Laning 和 Zierler 的“代数系统”将以熟悉的代数形式输入的命令转换为 Whirlwind 可以执行的机器代码。²⁷（术语上仍然存在一些歧义：虽然 Laning 和 Zierler 在其手册标题中使用了“翻译”一词，但在摘要中他们称之为“解释程序”。）²⁸

The first programming system to operate in the sense of a modern compiler was developed by J. H. Laning and N. Zierler for the Whirlwind computer at the Massachusetts Institute of Technology in the early 1950s. They described their system, which never had a name, in an elegant and terse manual entitled “A Program for Translation of Mathematical Equations for Whirlwind I,” distributed by MIT to about one-hundred locations in January 1954.²⁶ It was, in John Backus’s words, “an elegant concept elegantly realized.” Unlike the UNIVAC compilers, this system worked much as modern compilers work; that is, it took as its input commands entered by a user, and generated as output fresh and novel machine code, which not only executed those commands but also kept track of storage locations, handled repetitive loops, and did other housekeeping chores. Laning and Zierler’s “Algebraic System” took commands typed in familiar algebraic form and translated them into machine codes that Whirlwind could execute.²⁷ (There was still some ambiguity as to the terminology: while Laning and Zierler used the word “translate” in the title of their manual, in the Abstract they call it an “interpretive program.”)²⁸

不应过多地解读这个系统。它不是一种通用编程语言，而是一种求解代数方程的方法。Whirlwind 的用户并不特别关心 UNIVAC 客户感兴趣的商业应用。尽管 Backus 注意到了它的优雅，但他也指出，尽管当时 Whirlwind 备受关注，但它几乎被忽视了。29^在他看来，它之所以被忽视，是因为它威胁到了他所谓的程序员的“圣职”，程序员对自己使用很少有人能理解的技术和技巧处理机器代码的能力感到一种反常的自豪，这种态度一直持续到个人电脑时代。Donald Knuth 在 1980 年调查了早期的编程系统，他认为 Laning 和 Zierler 系统比 Whirlwind 的其他编码系统慢十倍的指控还有另一个原因。30^对于Knuth 来说，这句话来自一位描述过 MIT 使用的各种系统的人，这句话充满了谴责之词。³¹缩小自动编译器和手工编码之间的差距对于赢得编译器系统的认可和打破程序员的神职人员地位是必要的。

One should not read too much into this system. It was not a general-purpose programming language but a way of solving algebraic equations. Users of the Whirlwind were not particularly concerned with the business applications that interested UNIVAC customers. Although Backus noted its elegance, he also remarked that it was all but ignored, despite the publicity given Whirlwind at that time.²⁹ In his opinion, it was ignored because it threatened what he called the “priesthood” of programmers, who took a perverse pride in their ability to work in machine code using techniques and tricks that few others could fathom, an attitude that would persist well into the era of personal computers. Donald Knuth, who surveyed early programming systems in 1980, saw another reason in the allegation that the Laning and Zierler system was slower by a factor of ten than other coding systems for Whirlwind.³⁰ For Knuth, that statement, by someone who had described various systems used at MIT, contained damning words.³¹ Closing that gap between automatic compilers and hand coding would be necessary to win acceptance for compiler systems and to break the priesthood of the programmers.

汇编程序

Assemblers

这些系统最终得到了改进，并被称为编程语言。该术语的出现与它们与自然语言共享一些受限属性有关，例如语法规则。软件开发的历史通常与高级编程语言的历史同义——这些语言从更接近代数或典型用户描述过程的方式的代码生成机器代码。然而，尽管这些所谓的高级语言很重要，但直到 20 世纪 60 年代，许多设备的编程仍在较低级别上进行。虽然也被称为“语言”，但这些代码通常只为程序员在其中编码的每条指令生成一条或最多几条机器指令。每个代码都被一一翻译成计算机可以直接执行的相应二进制数。程序不是编译的，而是“汇编的”，执行此操作的程序称为“汇编程序”。这种一一对应关系有一些扩展，形式是与多条机器指令相对应的“宏”指令。一些商业设施维护着处理排序和合并操作的大型宏库；这些宏库与标准汇编语言指令相结合，构成了许多商业设施的软件开发，即使高级语言得到了改进。

These systems eventually were improved and came to be known as Programming Languages. The emergence of that term had to do with their sharing of a few restricted attributes with natural language, such as rules of syntax. The history of software development has often been synonymous with the history of high-level programming languages— languages that generated machine codes from codes that were much closer to algebra or to the way a typical user might describe a process. However, although these so-called high-level languages were important, programming at many installations continued to be done at much lower levels well into the 1960s. Though also called “languages,” these codes typically generated only a single, or at most a few, machine instructions for each instruction coded by a programmer in them. Each code was translated, one-to-one, into a corrresponding binary number that the computer could directly execute. A program was not compiled but “assembled,” and the program that did that was called an “assembler.” There were some extensions to this one-to-one correspondence, in the form of “macro” instructions that corresponded to more than one machine instruction. Some commercial installations maintained large libraries of macros that handled sorting and merging operations; these, combined with standard assembly-language instructions, comprised software development at many commercial installations, even as high-level languages improved.

典型的汇编程序命令可能是“LR”，后跟内存地址代码。汇编程序会将其转换为二进制数字，以执行“将某个内存位置的内容加载到中央处理器的寄存器中”的操作。一个重要特性是使用符号标签来表示内存位置，其数字机器地址可能会根据编写程序时未知的其他情况而改变。汇编程序的任务是当遇到变量符号时分配适当数量的机器存储空间，并在程序执行过程中跟踪此存储空间。IBM 计算机用户组 SHARE（下文将介绍）在开发 IBM 704 汇编程序方面发挥了作用，并且汇编语言一直受到 IBM 计算机 System/360 系列的大力支持。

A typical assembler command might be “LR” followed by a code for a memory address. The assembler would translate that into the binary digits for the operation “Load the contents of a certain memory location into a register in the central processor.” An important feature was the use of symbolic labels for memory locations, whose numerical machine address could change depending on other circumstances not known at the time the program was written. It was the job of the assembler program to allocate the proper amount of machine storage when it encountered a symbol for a variable, and to keep track of this storage through the execution of the program. The IBM computer user’s group SHARE (described next) had a role in developing an assembler for the IBM 704, and assembly language continued to find strong adherents right up into the System/360 line of IBM computers.

SHARE (1955)

当计算机供应商和设计人员都在研究高级语言时，规模虽小但不断增长的客户群体决定从另一个方向着手软件开发。1955 年，洛杉矶地区的一组 IBM 701 用户面临着将他们的设备升级到新的 IBM 704 的艰巨前景，他们联合起来，希望分享经验比单打独斗更好。那年 8 月，他们在圣莫尼卡兰德公司的中立地区会面。在兰德会面避免了因用户代表北美航空公司和洛克希德等竞争公司而产生的问题。这个自称为 SHARE ³²的团体发展迅速，很快就开发出了一个令人印象深刻的例程库，例如用于处理矩阵的例程库，每个成员都可以使用。

While computer suppliers and designers were working on high-level languages, the small but growing community of customers decided to tackle software development from the other direction. In 1955, a group of IBM 701 users located in the Los Angeles area, faced with the daunting prospect of upgrading their installations to the new IBM 704, banded together in the expectation that sharing experiences was better than going alone. That August they met on the neutral territory of the RAND Corporation in Santa Monica. Meeting at RAND avoided problems that stemmed from the fact that users represented competing companies like North American Aviation and Lockheed. Calling itself SHARE,³² the group grew rapidly and soon developed an impressive library of routines, for example, for handling matrices, that each member could use.

IBM 多年来一直赞助自己的制表设备客户支持，但 SHARE 的快速发展表明存储程序数字计算机的世界有多么不同。不到一年，会员人数（所有大型 IBM 系统的客户）就增长到 62 名。SHARE 的成立对 IBM 来说可能是一件好事，因为 SHARE 帮助加快了 IBM 设备的接受度，并可能帮助了 704 的销售。随着 SHARE 人数和实力的增长，它对 IBM 计算机和软件的未来发展方向形成了强烈的看法，IBM 别无选择，只能承认 SHARE 的地位。随着更小、更便宜的计算机出现在市场上，这些团体的价值和影响力将会增加。例如，数字设备小型计算机的用户组 DECUS 与 DEC 的关系非常密切，而对于个人计算机，用户组将变得更加重要，这将在第 7 章中讨论。

IBM had for years sponsored its own version of customer support for tabulator equipment, but the rapid growth of SHARE shows how different was the world of stored-program digital computers. Within a year the membership—all customers of large IBM systems—had grown to sixty-two members. The founding of SHARE was probably a blessing for IBM, since SHARE helped speed the acceptance of IBM’s equipment and probably helped sales of the 704. As SHARE grew in numbers and strength, it developed strong opinions about what future directions IBM computers and software ought to take, and IBM had little choice but to acknowledge SHARE’s place at the table. As smaller and cheaper computers appeared on the market, the value and clout of the groups would increase. For instance, DECUS, the users group for Digital Equipment minicomputers, had a very close relationship with DEC, and for personal computers the users groups would become even more critical, as will be discussed in chapter 7.

数据排序

Sorting Data

无论使用何种级别的编程语言，所有商业和许多科学设备都必须应对与硬件性质密切相关的活动，即处理称为文件的集合数据，这些文件由按顺序存储在磁带卷上的记录组成。尽管磁带比穿孔卡有许多优势，但它与卡片的相似之处在于它按顺序一个接一个地存储记录。为了使用这些数据，人们经常必须按顺序（例如字母顺序）对其进行排序，以便找到特定的记录。对数据（数字和非数字）进行排序在早期的商业计算中占主导地位，据估计，到 1973 年，这占据了所有计算机时间的 25%。33^在极端情况下，人们可能不得不在对一两个记录进行少量更改后对一个非常大的文件进行排序——这显然是一种低效且昂贵的计算机时间使用方式。建立公司数据处理系统的分析师通常会尽量减少这种情况，但他们无法完全避免。

Regardless of what level of programming language they used, all commercial and many scientific installations had to contend with an activity that was intimately related to the nature of the hardware— namely, the handling of data in aggregates called files, which consisted of records stored sequentially on reels of tape. Although tape offered many advantages over punched cards, it resembled cards in the way it stored records one after the other in a sequence. In order to use this data, one frequently had to sort it into an order (e.g., alphabetic) that would allow one to find a specific record. Sorting data (numeric as well as nonnumeric) dominated early commercial computing, and as late as 1973 was estimated to occupy 25 percent of all computer time.³³ In an extreme case, one might have to sort a very large file after having made only a few changes to one or two records—obviously an inefficient and costly use of computer time. Analysts who set up a company’s data processing system often tried to minimize such situations, but they could not avoid them entirely.

计算机编程与此类操作同步。公司会定期处理数据、更新文件并打印一系列报告。处理操作中有一个程序用于对文件进行排序并打印出按一个或多个键排序的各种报告。这些报告被打印出来并装订成文件夹，组织中的人员正是通过这些打印文件才能访问公司数据。例如，如果客户致电保险公司询问其账户情况，员工会参考最新的打印输出，可能按客户编号排序。因此，对记录进行排序并合并到已排序文件中是数据处理的主要方式，直到开发出允许直接访问特定记录的存储方法（例如磁盘）。随着这些方法的成熟，排序的需要逐渐减少，但并没有完全消失——事实上，一些最有效的排序方法是在这些变化发生的时候（20 世纪 60 年代末）发明的^。34

Computer programming was synchronized to this type of operation. On a regular basis a company’s data would be processed, files updated, and a set of reports printed. Among the processing operations was a program to sort a file and print out various reports sorted by one or more keys. These reports were printed and bound into folders, and it was from these printed volumes that people in an organization had access to corporate data. For example, if a customer called an insurance company with a question about his or her account, an employee would refer to the most recent printout, probably sorted by customer number. Therefore sorting and merging records into a sorted file dominated data processing, until storage methods (e.g., disks) were developed that allowed direct access to a specific record. As these methods matured, the need to sort diminished but did not go away entirely—indeed, some of the most efficient methods for sorting were invented around the time (late 1960s) that these changes were taking place.³⁴

如第 1 章所述，约翰·冯·诺依曼仔细评估了 EDVAC 的拟议设计，以确定其数据排序能力。他推断，如果 EDVAC 的排序能力与打孔卡排序机一样好，那么它就有资格成为一台万能机器。35 ¹⁹⁴⁵年，冯·诺依曼手写了一份 EDVAC 排序清单，被认为“可能是现存最早的存储程序计算机程序”。36^埃克特和莫奇利开始建造 UNIVAC 时承担的首要任务之一就是为其开发排序程序。

As mentioned in chapter 1, John von Neumann carefully evaluated the proposed design of the EDVAC for its ability to sort data. He reasoned that if the EDVAC could sort as well as punched-card sorting machines, it would qualify as an all-purpose machine.³⁵ A 1945 listing in von Neumann’s handwriting for sorting on the EDVAC is considered “probably the earliest extant program for a stored-program computer.”³⁶ One of the first tasks that Eckert and Mauchly took on when they began building the UNIVAC was to develop sorting routines for it.

实际上，他们聘请了其他人来做这件事，Frances E. (Betty) Holberton，她是跟随 Eckert 和 Mauchly 从 ENIAC 到 Eckert-Mauchly 计算机公司的人之一。Mauchly 让她负责开发 UNIVAC 软件（尽管当时还没有使用这个词）。37 ^Holberton的首批产品之一是 1952 年投入使用的程序，用于读取和排序存储在 UNIVAC 磁带驱动器上的数据。Donald Knuth 称其为“有史以来为自动编程开发的第一个主要‘软件’程序” ^。38

Actually, they hired someone else to do that, Frances E. (Betty) Holberton, one of the people who followed Eckert and Mauchly from the ENIAC to the Eckert–Mauchly Computer Corporation. Mauchly gave her the responsibility for developing UNIVAC software (although that word was not in use at the time).³⁷ One of Holberton’s first products, in use in 1952, was a routine that read and sorted data stored on UNIVAC tape drives. Donald Knuth called it “the first major ‘software’ routine ever developed for automatic programming.”³⁸

Holberton 在电子数据处理刚开始的那段日子里开发的技术确立了一种模式，这种模式被沿用了多年。在上述保险公司中，我们可以假设其客户记录已经按客户编号顺序放在磁带上。如果删除或更改了新帐户，计算机必须在磁带上找到该帐户的正确位置，进行更改或删除，并将剩余帐户重新排列到磁带上的其他位置。因此，一个简单的更改可能会涉及移动大量数据。更实际的做法是只更改整个文件的一小部分，同时添加和删除几个帐户。这些记录将被排序并写入单个磁带卷上的一个小文件；然后，通过将记录插入主文件上的适当位置，将该文件合并到主文件中，就像桥牌玩家在发牌时将牌插入手中一样。因此，每次运行都会创建一个新的“主”文件，而之前的主文件则作为备份保留，以防机械故障^。39

The techniques Holberton developed in those first days of electronic data processing set a pattern that was followed for years. In the insurance company mentioned above, we can assume that its customer records have already been placed on the tape sequentially in customer number order. If a new account was removed or changed, the computer had to find the proper place on the tape where the account was, make the changes or deletions, and shuffle the remaining accounts onto other positions on the tape. A simple change might therefore involve moving a lot of data. A more practical action was to make changes to a small percentage of the whole file, adding and deleting a few accounts at the same time. These records would be sorted and written to a small file on a single reel of tape; then this file would be merged into the master file by inserting records into the appropriate places on the main file, like a bridge player inserting cards into his or her hand as they are dealt. Thus for each run a new “master” file was created, with the previous master kept as a backup in case of a mechanical failure.³⁹

由于磁带中存储的记录数量远远超过计算机内存所能容纳的数量，因此程序必须将磁带中的小块记录读入主内存，在内部对它们进行排序，将排序后的块写入磁带，然后获取下一个块。排序后的块被合并到磁带上的主文件中，直到整个文件处理完毕。至少同时使用两个磁带驱动器，磁带既向前读，又向后读。Holberton 和她在 UNIVAC 的团队开发的程序是管理复杂性的杰作；但即使在编写这些程序时，她也意识到，如果可以组织问题，使其能够在不诉诸大量排序的情况下解决问题，那就更好了。40^随着磁盘存储和“链表”概念的出现，其中列表中的每个记录都包含有关下一个（或上一个）记录的位置的信息，排序失去了主导地位。

Because the tape held far more records than could fit in the computer’s internal memory, the routines had to read small blocks of records from the tape into the main memory, sort them internally, write the sorted block onto a tape, and then fetch the next block. Sorted blocks were merged onto the master file on the tape until the whole file was processed. At least two tape drives were used simultaneously, and the tapes were read both forward and backward. The routines developed by Holberton and her team at UNIVAC were masterpieces of managed complexity; but even as she wrote them she recognized that it would be better if one could organize a problem so that it could be solved without recourse to massive sorts.⁴⁰ With the advent of disk storage and a concept of “linked lists,” in which each record in a list contained information about where the next (or previous) record was, sorting lost its dominance.

FORTRAN（1957年）

FORTRAN (1957)

1957 年初，IBM 为 704 计算机引入了编程语言 FORTRAN（“公式翻译”——首选拼写为大写）。从一开始，该语言就在 IBM 客户中大获成功，而且经过多次修改和扩展，该语言仍被广泛使用。41 ^FORTRAN的成功有很多因素。其一是它的语法（符号的选择和使用规则）与普通代数非常接近，主要区别在于在穿孔卡上标上标或下标的难度。工程师喜欢它的熟悉性；他们也喜欢简洁明了、易于阅读的用户手册。也许最重要的因素是它避免了 Laning 和 Zierler 系统的速度损失。FORTRAN 编译器生成的机器代码与人类编写的代码一样高效和快速。John Backus 强调了这一点，尽管批评者指出 FORTRAN 在高级语言中并不是独一无二的。⁴² IBM 的市场主导地位显然也对 FORTRAN 的成功起到了一定作用，但如果 704 型不是一台功能强大且设计精良的计算机来运行 FORTRAN，IBM 的优势就不会持续下去。巴克斯还指出，704 型硬件中浮点运算的提供促使他开发了一种高效、快速的 FORTRAN 编译器，因为不再有任何繁琐而缓慢的浮点例程可以“隐藏”在后面^。43

The programming language FORTRAN (“Formula Translation”—the preferred spelling was all capitals) was introduced by IBM for the 704 computer in early 1957. It was a success among IBM customers from the beginning, and the language—much modified and extended— continues to be widely used.⁴¹ Many factors contributed to the success of FORTRAN. One was that its syntax—the choice of symbols and the rules for using them—was very close to what ordinary algebra looked like, with the main difference arising from the difficulty of indicating superscripts or subscripts on punched cards. Engineers liked its familiarity; they also liked the clear, concise, and easy-to-read users manual. Perhaps the most important factor was that it escaped the speed penalty incurred by Laning and Zierler’s system. The FORTRAN compiler generated machine code that was as efficient and fast as code written by human beings. John Backus emphasized this point, although critics have pointed out that FORTRAN was not unique among high-level languages.⁴² IBM’s dominant market position obviously also played a role in FORTRAN’s success, but IBM’s advantage would not have endured had the Model 704 not been a powerful and well-designed computer on which to run FORTRAN. Backus also noted that the provision, in the 704’s hardware, of floating-point arithmetic drove him to develop an efficient and fast compiler for FORTRAN, as there were no longer any cumbersome and slow floating-point routines to “hide” behind.⁴³

FORTRAN 最初的成功表明，用户多么乐意接受一个隐藏机器内部工作细节的系统，让他们可以专心解决自己的问题，而不是机器的问题。同时，在 20 世纪 90 年代，当新的语言隐藏了更多复杂性时，FORTRAN 仍在继续使用，这暴露了这种理念的局限性。C 语言是由贝尔实验室开发的，也是 1980 年后最流行的语言之一，它与 FORTRAN 一样，允许程序员在需要时访问低级操作。成功且经久不衰的计算机语言（其中为数不多）似乎都具有这种特性，即向程序员隐藏部分（但不是全部）计算机内部工作。

FORTRAN’s initial success illustrates how readily users embraced a system that hid the details of the machine’s inner workings, leaving them free to concentrate on solving their own, not the machine’s, problems. At the same time, its continued use into the 1990s, at a time when newer languages that hide many more layers of complexity are available, reveals the limits of this philosophy. The C language, developed at Bell Labs and one of the most popular after 1980, shares with FORTRAN the quality of allowing a programmer access to low-level operations when that is desired. The successful and long-lasting computer languages, of which there are very few, all seem to share this quality of hiding some, but not all, of a computer’s inner workings from its programmers.

柯博

COBOL

FORTRAN 的成功与几年后开发的 COBOL（“通用商业导向语言”）在商业领域的成功不相上下。COBOL 的成功归功于美国国防部，该部门于 1959 年 5 月召集了一个委员会来解决开发通用商业语言的问题；会议之后进行了短暂而集中的努力来制定该语言的规范，并于当年年底发布了初步规范。这些规范一经发布，几家制造商就开始为各自的计算机编写编译器。第二年，美国政府宣布不会购买或租赁无法处理 COBOL 的计算机设备。44^因此，COBOL 成为首批标准化的语言之一，标准化到同一程序可以在不同供应商的不同计算机上运行并产生相同结果的程度。这一里程碑的首次记录发生在 1960 年 12 月，当时同一程序（略有改动）在 UNIVAC II 和 RCA 501 上运行。然而，COBOL 是否设计精良且功能强大仍有争议。

FORTRAN’s success was matched in the commercial world by COBOL (“Common Business Oriented Language”), developed a few years later. COBOL owed its success to the U.S. Department of Defense, which in May 1959 convened a committee to address the question of developing a common business language; that meeting was followed by a brief and concentrated effort to produce specifications for the language, with preliminary specifications released by the end of that year. As soon as those were published, several manufacturers set out to write compilers for their respective computers. The next year the U.S. government announced that it would not purchase or lease computer equipment that could not handle COBOL.⁴⁴ As a result, COBOL became one of the first languages to be standardized to a point where the same program could run on different computers from different vendors and produce the same results. The first recorded instance of that milestone occurred in December 1960, when the same program (with a few minor changes) ran on a UNIVAC II and an RCA 501. Whether COBOL was well designed and capable is still a matter of debate, however.

COBOL 的部分起源可以追溯到 Grace Hopper 在 UNIVAC 编译器上的工作。到 1956 年，她开发了一种名为“B-0”的编译器，在某些版本中也被称为“MATH-MATIC”或“FLOW-MATIC”，与她的“A”系列编译器不同，它面向商业应用。IBM 的一个名为 Commercial Translator 的项目也产生了一定的影响。通过与新成立的计算机科学公司签订合同，霍尼韦尔还开发了一种许多人认为比 COBOL 更好的语言，但结果“FACT”并未获得美国政府的认可。尽管如此，FACT 对后来的 COBOL 开发产生了影响；它的遗产之一是它在创办计算机科学公司（首批商业软件公司之一）方面发挥的作用^。45

Part of COBOL’s ancestry can be traced to Grace Hopper’s work on the compilers for the UNIVAC. By 1956 she had developed a compiler called “B-0,” also called in some incarnations “MATH-MATIC” or “FLOW-MATIC,” which unlike her “A” series of compilers was geared toward business applications. An IBM project called Commercial Translator also had some influence. Through a contract with the newly formed Computer Sciences Corporation, Honeywell also developed a language that many felt was better than COBOL, but the result, “FACT,” did not carry the imprimatur of the U.S. government. FACT nevertheless had an influence on later COBOL development; part of its legacy was its role in launching Computer Sciences Corporation, one of the first commercial software companies.⁴⁵

正是从 Grace Hopper 那里，COBOL 获得了其最著名的属性，即能够使用长字符名称，从而使生成的语言看起来像普通英语。例如，在 FORTRAN 中，人们可能会写：

It was from Grace Hopper that COBOL acquired its most famous attribute, namely, the ability to use long character names that made the resulting language look like ordinary English. For example, whereas in FORTRAN one might write:

如果 A > B

IF A > B

相应的 COBOL 语句可能是：

the corresponding COBOL statement might read:

如果员工工作时间大于最高^{46 小时}

IF EMPLOYEE-HOURS IS GREATER THAN MAXIMUM⁴⁶

支持者认为，这种设计使 COBOL 更易于阅读和理解，尤其是对于使用该程序但与编写程序关系不大的“管理人员”而言。支持者还认为，这使得程序具有“自文档化”功能：程序员无需在列表中插入注释（即编译器忽略但人类可以阅读和理解的描述）。对于 COBOL，实际的指令列表对于人类和机器来说都是足够好的描述。后来人们已经知道了一件事情：在编写代码几个月后，即使是编写者，更不用说其他人，也无法说出该代码应该做什么。

Proponents argued that this design made COBOL easier to read and understand, especially by “managers” who used the program but had little to do with writing it. Proponents also argued that this made the program “self-documenting”: programmers did not need to insert comments into the listing (i.e., descriptions that were ignored by the compiler but that humans could read and understand). With COBOL, the actual listing of instructions was a good enough description for humans as well as for the machine. It was already becoming known what later on became obvious: a few months after a code was written, even the writer, never mind anyone else, cannot tell what that code was supposed to do.

与 FORTRAN 一样，COBOL 也幸存下来，甚至在个人电脑时代蓬勃发展。然而，其类似英语的语法并没有取得其创造者所希望的成功。许多程序员对变量的神秘代码感到满意，他们很少使用将变量描述为较长的英语单词的能力。无论如何，并非所有经理都觉得这种语言易于阅读。尽管如此，它还是提供了一些文档，这总比没有好——太多程序都是没有文档的。在随后的几年里，研究人员探索了机器语言和人类语言之间的关系，虽然 COBOL 是一个重要的里程碑，但它给人一种错觉，认为它比实际更了解英语。让计算机“按照我的意思做，而不是按照我说的做”仍然是计算机科学研究的前沿。

Like FORTRAN, COBOL survived and even thrived into the personal computer era. Its English-like syntax did not achieve the success its creators hoped for, however. Many programmers felt comfortable with cryptic codes for variables, and they made little use of the ability to describe variables as longer English words. Not all managers found the language easy to read anyway. Still, it provided some documentation, which was better than none—and too many programs were written with none. In the years that followed, researchers explored the relationship between machine and human language, and while COBOL was a significant milestone, it gave the illusion that it understood English better than it really did. Getting computers to “do what I meant, not what I said” is still at the forefront of computer science research.

2001 年已经过去了，但计算机语言远不及 HAL 所展现的自然语言理解水平，而 HAL 正是斯坦利·库布里克 (Stanley Kubrick) 的电影《2001：太空漫游》中的主角（图 3.3）。47^随着2000 年的临近，工业界面临着一个更为严重的问题：许多计算机程序无法识别当年份用两位十进制数字表示时，如果第一位是零，那么它表示的是 2000、2001 等，而不是 1900、1901 等。许多有问题的程序都是在 20 世纪 60 年代用 COBOL 编写的。为了修复这些问题，熟悉 COBOL 的程序员不得不仔细查阅旧的程序列表，找到并更正有问题的代码。这个故事在互联网讨论组中流传开来，这些讨论组的公司访问了佛罗里达州的退休社区，并劝说老 COBOL 程序员离开高尔夫球场。 “Year2000 Bug” 充分证明了，尽管可以编写自文档化的 COBOL 程序，但很少有人这样做。需要更正的程序对于许多现代软件开发领域的最佳实践者来说都是难以理解的。

The year 2001 came and went, with computer languages coming nowhere near the level of natural language understanding shown by HAL, the computer that was the star of the Stanley Kubrick movie 2001: A Space Odyssey (figure 3.3).⁴⁷ As the year 2000 approached, the industrial world faced a more serious issue: the inability of many computer programs to recognize that when a year is indicated by two decimal digits, the first of which is a zero, it means 2000, 2001, etc., not 1900, 1901, etc. Many of those offending programs were written in the 1960s, in COBOL. In order to fix them, programmers familiar with COBOL had to wade through old program listings and find and correct the offending code. A story circulated around Internet discussion groups of companies visiting retirement communities in Florida and coaxing old-time COBOL programmers off the golf course. The “Year2000 Bug” gave ample evidence that, although it was possible to write COBOL programs that were self-documenting, few ever did. The programs that had to be corrected were incomprehensible to many of the best practitioners of modern software development.

“语言”这个词原来是一个危险的术语，它所暗示的含义远远超出了它的最初使用者的预见。这个英语单词源于法语langue，意为舌头，意味着它是说出来的。无论它与自然语言有哪些相似之处，计算机语言不是说出来的，而是根据严格定义和精确的语法书写出来的。48^霍珀曾经讲述过她如何开发一个版本的 FLOW-MATIC，在这个版本中，她用法语替换了所有英语术语，例如“输入”、“写入”等。当她向 UNIVAC 的一位高管展示这个版本时，她被立即赶出了办公室。后来她意识到，计算机的概念本身对这位高管来说就是一个威胁；让它“说”法语——一种他不会说的语言——太过分了^。49

The word “language” turned out to be a dangerous term, implying much more than its initial users foresaw. The English word is derived from the French langue, meaning tongue, implying that it is spoken. Whatever other parallels there may be with natural language, computer languages are not spoken but written, according to a rigidly defined and precise syntax.⁴⁸ Hopper once recounted how she developed a version of FLOW-MATIC in which she replaced all the English terms, such as “Input,” “Write,” and so on, with their French equivalents. When she showed this to a UNIVAC executive, she was summarily thrown out of his office. Later on she realized that the very notion of a computer was threatening to this executive; to have it “speaking” French—a language he did not speak—was too much.⁴⁹

图 3.3《2001：太空漫游》

中的一幕。两名宇航员之间可以看到机载计算机 HAL 的摄像镜头。1968 年，伴随该电影上映的宣传称，该电影的创作者描绘的技术水平在 2001 年是先进的，但并非不合理。计算机变得越来越强大、越来越紧凑，但事实证明，到 2001 年，机器对自然语言的理解还达不到 HAL 所展示的水平。（来源：《2001：太空漫游》。© 1968 Turner Entertainment Co.）

Figure 3.3

A scene from 2001: A Space Odyssey. A camera eye of HAL, the on-board computer, is visible between the two astronauts. Publicity that accompanied the film’s release in 1968 stated that its creators depicted a level of technology that they felt was advanced but not unreasonably so for the year 2001. Computers have become more powerful and more compact, but it turned out that machine understanding of natural language, on a level shown by HAL, was not attained by 2001. (Source : 2001: A Space Odyssey. © 1968 Turner Entertainment Co.)

语言与软件

Languages Versus Software

我们可以以 FORTRAN 和 COBOL 这两个高峰为视角纵观整个 20 世纪 60 年代的软件领域。在认识到汇编语言的重要地位，然后再回顾其他几种高级语言的历史之后，我们或许可以得出一个完整的结论。

From the twin peaks of FORTRAN and COBOL we can survey the field of software through the 1960s. After recognizing the important place of assembly language and then looking at the histories of a few more high-level languages, we might conclude that we have a complete picture.

高级语言中包括 ALGOL，它主要在欧洲开发，开发时间在 1958 年至 1960 年间，被提议作为 FORTRAN 的更严格的替代方案。ALGOL 从一开始就旨在独立于任何特定的硬件配置，这与最初的 FORTRAN 不同，后者的命令指向 IBM 704 处理器的特定寄存器。与 FORTRAN 不同的是，ALGOL 的定义非常谨慎且形式化，因此用它编写的表达式的作用不会产生歧义。该定义本身在一种称为“BNF”（Backus-Normal-Form 或 Backus-Naur-Form）的语言中指定。它的欧洲贡献者们对 ALGOL 寄予厚望，希望它能成为不受 IBM 约束的全球标准，但这并没有发生。ALGOL 委员会的一名成员遗憾地指出，ALGOL 这个名字是算法语言的缩写，也是一颗星星的名字，其英文翻译为“食尸鬼”。无论其命运如何，ALGOL 仍然对后来的语言产生了影响。

Among the high-level languages was ALGOL, developed mainly in Europe between 1958 and 1960 and proposed as a more rigorous alternative to FORTRAN. ALGOL was intended from the start to be independent of any particular hardware configuration, unlike the original FORTRAN with its commands that pointed to specific registers of the IBM 704’s processor. Also unlike FORTRAN, ALGOL was carefully and formally defined, so that there was no ambiguity about what an expression written in it would do. That definition was itself specified in a language known as “BNF” (Backus-Normal-Form or Backus-Naur-Form). Hopes were high among its European contributors that ALGOL would become a worldwide standard not tied to IBM, but that did not happen. One member of the ALGOL committee ruefully noted that the the name ALGOL, a contraction of Algorithmic Language, was also the name of a star whose English translation was “the Ghoul.” Whatever the reason for its ill fate, ALGOL nonetheless was influential on later languages.

在当时开发的许多其他语言中，只有少数几种广为人知，没有一种能像 FORTRAN 或 COBOL 那样成功。JOVIAL（Jules [Schwartz] 自己的国际代数语言版本）是美国国防部为 SAGE 防空系统开发的 ALGOL 的一个变体；它仍然用于防空和空中交通管制应用。LISP（列表处理）是一种面向处理符号而不是评估代数表达式的语言；它一直是人工智能研究人员最喜欢的语言。SNOBOL（面向字符串的符号语言）面向处理“字符串”——字符序列，如文本。20 世纪 60 年代后期开发的其他几种语言将在后面讨论^。50

Of the many other languages developed at this time, only a few became well known, and none enjoyed the success of FORTRAN or COBOL. JOVIAL (Jules [Schwartz’s] Own Verison of the International Algebraic Language) was a variant of ALGOL developed by the Defense Department, in connection with the SAGE air-defense system; it is still used for air-defense and air-traffic-control applications. LISP (List Processing) was a language oriented toward processing symbols rather than evaluating algebraic expressions; it has been a favorite language for researchers in artificial intelligence. SNOBOL (StriNg-Oriented symBOlic Language) was oriented toward handling “strings”— sequences of characters, like text. A few of the many other languages developed in the late 1960s will be discussed later.⁵⁰

在汇编程序和 COBOL 之间有一种用于 IBM 计算机的系统，称为 RPG（报告程序生成器；其他制造商也有类似的系统，只是名称不同）。整个 20 世纪 60 年代及之后，这些系统在许多商业安装中得到广泛使用。编程教科书没有将 RPG 归类为一种语言，但在某些方面，它比 COBOL 运行在更高的级别。51 ^RPG类似于填写预先打印的表格。程序员不必指定计算机要对数据执行哪些操作。而是根据数据在表格上输入的位置来指定操作（例如，像在所得税申报表上一样）。显然，RPG 只适用于日常不会发生变化的常规、结构化问题，但在这些情况下，它使程序员从很多细节中解放出来。它仍然用于常规文书操作。

Somewhere between assemblers and COBOL was a system for IBM computers called RPG (Report Program Generator; other manufacturers had similar systems with different names). These were in common use in many commercial installations throughout the 1960s and after. Textbooks on programming do not classify RPG as a language, yet in some ways it operated at a higher level than COBOL.⁵¹ RPG was akin to filling out a preprinted form. The programmer did not have to specify what operations the computer was to perform on the data. Instead the operations were specified by virtue of where on the form the data were entered (e.g., like on an income tax return). Obviously RPG worked only on routine, structured problems that did not vary from day to day, but in those situations it freed the programmer from a lot of detail. It is still used for routine clerical operations.

系统软件

System Software

除了解决用户问题的程序外，还出现了一组不同的程序，它们的作用是分配计算机系统处理单个任务时的资源。即使是准备工资单等最常规的工作也可能有很多变化，而执行此操作的程序可能会与以不同方式使用文件的短得多的程序交织在一起。一个程序可能只包含几行代码并涉及几段数据；另一个程序可能使用一些记录，但对它们进行大量处理；第三个程序可能使用所有记录，但处理较少，等等。

Besides programs that solved a user’s problem, there arose a different set of programs whose job was to allocate the resources of the computer system as it handled individual jobs. Even the most routine work of preparing something like a payroll might have a lot of variation, and the program to do that might be interspersed with much shorter programs that used the files in different ways. One program might consist of only a few lines of code and involve a few pieces of data; another might use a few records but do a lot of processing with them; a third might use all the records but process them less, and so on.

早期的安装依靠操作员的判断来安排这些作业。随着问题的数量和复杂性不断增加，人们开始开发程序，到 20 世纪 90 年代，这些程序将主导整个行业。这些程序被称为操作系统（图 3.4）。最具创新性的早期工作是由用户完成的。通用汽车研究实验室于 1956 年开始设计的一个早期系统尤其具有影响力。52^它的成功帮助建立了批处理计算 - 将作业分组到一副卡片中，用控制卡隔开，以便正确设置机器，而人工操作员只需做很少的工作。这些系统的一个简单但关键的元素是使用特殊的“控制”卡，在保留的列中打入特定代码。这些代码告诉计算机接下来的卡片是 FORTRAN 程序或数据，或者新作业正在开始，等等。这演变成 IBM 称为作业控制语言 (JCL) 的系统。许多新手程序员对 JCL 卡有着清晰的记忆，其独特的双斜线 (//) 或斜线星号 (/*) 被打入某些字段。许多人还记得如果丢失一张卡会导致计算机将程序卡读取为数据，或反之亦然，从而导致混乱^。53

Early installations relied on the judgment of the operator to schedule these jobs. As problems grew in number and complexity, people began developing programs that, by the 1990s, would dominate the industry. These programs became known as operating systems (figure 3.4). The most innovative early work was done by users. One early system, designed at the General Motors Research Laboratories beginning in 1956 was especially influential.⁵² Its success helped establish batch computing—the grouping of jobs into a single deck of cards, separated by control cards that set the machine up properly, with only minimal work done by the human operator. A simple but key element of these systems was their use of special “control” cards with specific codes punched into reserved columns. These codes told the computer that the cards that followed were a FORTRAN program, or data, or that a new job was starting, and so on. That evolved into a system known at IBM as Job Control Language (JCL). Many a novice programmer has a vivid memory of JCL cards, with their distinctive double slash (//) or slash-asterisk (/*) punched into certain fields. Many also remember the confusion that resulted if a missing card caused the computer to read a program deck as data, or vice versa.⁵³

疯狂的

MAD

在大学环境中，也出现了类似高效管理工作流程的需求。学生课程往往在每周或每个学生之间都不一致，因此，让学生清楚地了解他们犯了哪些错误非常重要。（事实上，每个安装都需要这种清晰度，但当时很少有人意识到这一点。）1959 年，密歇根大学的 Bernie Galler、Bob Graham 和 Bruce Arden 开发了一个名为 MAD（密歇根算法解码器）的系统。MAD 基于 ALGOL，但与 ALGOL 不同的是，它以其他语言无法做到的方式处理了运行作业的细节。MAD 提供了快速编译，这对于教学环境至关重要，并且它具有良好的诊断功能，可帮助学生查找和纠正错误。这些品质使该系统不仅在教学方面取得了成功，而且在密歇根校园的物理学家、行为科学家和其他研究人员中也取得了成功。 MAD 的一个功能可能有助于它赢得学生的认可，那就是它在许多错误情况下打印出《疯狂杂志》吉祥物 Alfred E. Newman 的粗糙图片。（将图像编码到穿孔卡片上的 Bob Rosin 回忆说，这个“功能”最终被删除，因为学生们为了得到打印件故意犯错。）⁵⁴

In university environments there arose a similar need to manage the workflow efficiently. Student programs tended not to be uniform from week to week, or from one student to another, and it was important that students received clear messages about what kinds of errors they made. (In fact, every installation needed this clarity but few recognized that at the time.) In 1959 a system called MAD (Michigan Algorithmic Decoder) was developed at the University of Michigan by Bernie Galler, Bob Graham, and Bruce Arden. MAD was based on ALGOL, but unlike ALGOL it took care of the details of running a job in ways that few other languages could do. MAD offered fast compilation, essential for a teaching environment, and it had good diagnostics to help students find and correct errors. These qualities made the system not only successful for teaching, but also for physicists, behavioral scientists, and other researchers on the Michigan campus. One feature of MAD that may have helped win its acceptance among students was that it printed out a crude picture of Alfred E. Newman, the mascot of Mad Magazine, under many error conditions. (Bob Rosin, who coded the image on a deck of punched cards, recalled that this “feature” had eventually to be removed because students were deliberately making errors in order to get the printout.)⁵⁴

图 3.4

操作系统的起源和早期演进。（ a）单个用户的简单情况，可使用整个计算机的资源。请注意，该过程至少需要通过计算机两次：第一次将程序编译成机器语言，第二次加载并执行目标代码（可能打入新的卡片或存储在磁带卷上）。如果原始程序包含错误，计算机通常（但并非总是）会打印诊断消息，也可能打印“转储”，而不是尝试生成目标代码。（ b）如果用户以后想用不同的数据运行同一个程序，则无需重新编译原始程序。

Figure 3.4

The origins and early evolution of operating systems. (a) Simple case of a single user, with entire computer’s resources available. Note that the process requires a minimum of two passes through the computer: the first to compile the program into machine language, the second to load and execute the object code (which may be punched onto a new deck of cards or stored on a reel of tape). If the original program contained errors, the computer would usually, though not always, print a diagnostic message, and probably also a “dump,” rather than attempt to generate object code. (b) If at a later date the user wants to run the same program with different data, there is no need to recompile the original program.

图 3.4（续）

操作系统的起源和早期发展。（ c）“加载并运行”：密歇根算法解码器 (MAD) 系统折叠了目标代码的生成和执行，以便学生和其他用户可以更快地获得结果，或者在程序出现错误时获得诊断消息。

Figure 3.4(Continued)

The origins and early evolution of operating systems. (c) “Load and Go”: The Michigan Algorithmic Decoder (MAD) system collapsed the generation of object code and execution, so that students and other users could more quickly get results, or diagnostic messages if there were errors in their programs.

工业和教学设备的目标都是容纳不同长度和复杂程度的程序。出于经济原因，另一个目标是让计算机始终保持忙碌状态。不幸的是，很少有工业和商业设备像 MAD 的创建者那样意识到良好错误诊断的重要性。由于商业系统没有这样的诊断，许多教学环境也没有，他们认为学生迟早要习惯这个事实。在这些批处理系统中，如果程序包含哪怕一个简单的语法错误，操作系统都会决定计算机是否应该继续尝试解决问题。如果它决定不继续，它只会将相关内存部分的内容传输到打印机，将这些内容打印为数字行（甚至不将数字转换为十进制），暂停该程序的工作，然后继续执行队列中的下一个程序。这个过程的单词是“dump”。韦氏词典的定义是“粗暴地扔下或扔出”，这是恰当的。收到“核心转储”的倒霉用户实际上是在被粗鲁地告知，计算机认为其他人的工作更重要。试图根据核心转储打印的一排排数字来查找错误对于普通人来说是令人生畏的，甚至对于专家来说也是困难的。

Both industrial and teaching installations had the same goal of accommodating programs of different lengths and complexity. For economic reasons, another goal was to keep the computer busy at all times. Unfortunately, few industrial and commercial installations realized, as MAD’s creators did, the importance of good error diagnosis. And since the commercial systems did not have such diagnostics, many teaching environments did not, either, reasoning that students would sooner or later have to get used to that fact. In these batch systems, if a program contained even a simple syntax error, the operating system decided whether the computer should continue trying to solve the problem. If it decided not to, it would simply transfer the contents of the relevant portion of memory to a printer, print those contents as rows of numbers (not even translating the numbers into decimal), suspend work on that program, and go on to the next program in the queue. The word for that process was “dump.” Webster’s definition, “to throw down or out roughly,” was appropriate. The hapless user who received a “core dump” was in effect being told, rather rudely, that the computer had decided that someone else’s work was more important. Trying to find the error based on row upon row of numbers printed by a core dump was intimidating to the lay person and difficult even for the expert.

图 3.4（续）

操作系统的起源和早期发展。（ d）批处理。大型计算系统的经济性使得单个用户不可能像（ a）中那样独占机器。实际上，他或她的一副牌会与其他用户的“批处理”。程序和数据将由专门打孔的卡片分开；同样，每个人的工作将通过一张或多张“作业控制”卡与下一个人的工作分开。操作系统会根据每个用户的需要将适当的编译器加载到计算机中；它还可以从一副牌中提取数据并将其加载到更快的介质（如磁带或磁盘）上，并处理数据和结果（包括错误消息）的读取和打印。计算机操作员可能需要根据控制台的信号找到磁带并将其安装到驱动器上；操作员还会从打印机中取出打印输出并将其分开分发给每个用户。

Figure 3.4 (Continued)

The origins and early evolution of operating systems. (d) Batch processing. The economics of a large computing system made it unlikely that a single user could have exclusive use of the machine, as in (a). In practice his or her deck of cards would be “batched” with other users. The program and data would be separated by a specially punched card; likewise, each person’s job would be separated from the next person’s job by one or more “job control” cards. The operating system would load the appropriate compiler into the computer as each user required it; it might also extract the data from the deck of cards and load it onto a faster medium such as tape or disk, and handle the reading and printing of data and results, including error messages. The computer operator might be required to find and mount tapes onto drives, as indicated by a signal from the console; the operator would also pull the printout from the printer and separate it for distribution to each user.

图 3.4（续）

操作系统的起源和早期发展。（ e ）多道程序设计。如图 ( d )所示的混合程序将以不同的组合使用计算机的不同部分。一个程序可能大量使用磁带驱动器，但很少使用中央处理器的高级数学能力。因此，操作系统发展到支持“多道程序设计”：计算机能够同时运行多个程序，每个程序都使用另一个程序在特定时刻未使用的机器部分。操作系统会确保程序之间不会互相干扰，例如两个程序试图同时将数据写入同一内存位置。

Figure 3.4 (Continued)

The origins and early evolution of operating systems. (e) Multiprogramming. A mix of programs as in (d) would use different portions of the computer in a different mix. One program might make heavy use of the tape drives but little use of the central processor’s advanced mathematical powers. Operating systems thus evolved to support “multiprogramming”: the ability of a computer to run more than one program at the same time, each using portions of the machine that the other was not using at a given instant. The operating system took care that the programs did not interfere with one another, as in two programs attempting to write data at the same time to the same memory location.

随着操作系统的发展，它们往往会消耗更多宝贵的内存，直到几乎没有剩余内存来运行计算机最初安装的程序。操作系统名称的演变反映了它们日益增长的复杂性：它们被称为“监视器”，然后是“监控系统”，最后是“操作系统”。早期，客户开发了简单而精简的系统，例如 IBM 7090 系列的 Fortran 监控系统。事实证明，扩展到更复杂的系统很困难。SHARE 为 IBM 大型机开发的 SOS（Share 操作系统）更复杂，但效率更低。当 IBM 决定将其科学和商业系统系列合并为一个名为 System/360 的系列时，该公司还着手开发与之配套的操作系统 OS/360。55 ^System /360 对 IBM 来说是一个成功，并重新定义了该行业，正如后续章节将展示的那样，但 1966 年推出的操作系统 OS/360 却是一个失败，它的麻烦几乎让公司陷入困境。⁵⁶曾负责该项目的 Fred Brooks 写了一本关于 OS/360 开发的书，《人月神话》，这本书已成为管理大型软件项目难度的经典描述。Brooks 的许多见解之一是，委员会为软件项目制定了糟糕的结构；这也是上述共享操作系统以及 PL/I 和 ALGOL-68 语言（稍后讨论）出现问题的一个因素。

As operating systems evolved, they tended to consume more precious amounts of memory, until there was little left for running the programs the computer was installed for in the first place. The evolution of the name given to them reflects their growing complexity: They were called “monitors,” then “supervisor systems,” and finally “operating systems.” In the early days, simple and lean systems were developed by customers, for example, the Fortran Monitor System for the IBM 7090 series. Scaling up to more complex systems proved difficult. SOS (Share Operating System), developed by SHARE for IBM mainframes, was more complex but less efficient. When IBM decided to combine its line of scientific and business systems into a series called System/360, the company also set out to develop an operating system, OS/360, to go with it.⁵⁵ System/360 was a success for IBM and redefined the industry, as subsequent chapters will show, but the operating system OS/360, available by 1966, was a failure and its troubles almost sank the company.⁵⁶ Fred Brooks, who had been in charge of it, wrote a book on OS/360’s development, The Mythical Man-Month, which has become a classic description of the difficulty of managing large software projects. Among Brooks’s many insights is that committees make poor structures for software projects; this was also a factor in the problems with the Share Operating System noted above, as well as with the languages PL/I and ALGOL-68 (discussed later).

IBM 最终为其 360 系列开发了可行的系统软件，但当 20 世纪 60 年代中期发明小型计算机时，操作系统的历史又重新开始了：第一批小型计算机没有内部内存容量，只能支持简单的监视器，个人编写了运行良好的 Spartan 监视器。随着小型计算机的功能越来越强大，它们的操作系统也越来越强大，最终导致数字设备公司为其 VAX 系列计算机开发了 VMS（1978 年）。个人计算机再次出现了这种现象。第一批个人计算机具有基本的监视器，可以从磁带盒中加载数据，随后出现了更复杂但仍然精简的磁盘操作系统。最后，出现了基于“windows”的系统，其复杂性要求程序员团队对各个部分进行工作。正如预期的那样，其中一些项目伴随着 Brooks 在 System/360 中讨论的相同管理问题。计算机似乎被诅咒了，每隔十年左右就要经历这种痛苦的轮回。

IBM eventually developed workable system software for its 360 Series, but when the minicomputer was invented in the mid 1960s, the history of operating systems started over again: the first minis did not have the internal memory capacity to do more than support a simple monitor, and individuals wrote Spartan monitors that worked well. As minicomputers got more powerful, so did their operating systems, culminating in Digital Equipment Corporation’s VMS for its VAX line of computers (1978). The phenomenon was repeated yet again with the personal computer. The first personal computers had rudimentary monitors that loaded data from tape cassettes, and these were followed by more complex but still lean disk operating systems. Finally, “windows”-based systems appeared, whose complexity required teams of programmers working on subsections. As expected, some of these projects have been accompanied by the same management problems discussed by Brooks for the System/360. Computers seem to be cursed with having to go through this painful wheel of reincarnation every other decade or so.

计算机科学

Computer Science

这些例子表明，软件不仅仅是编程语言的开发和演化。但编程语言却主导了这一时期出现的学科，即计算机科学。该学科最早出现在 20 世纪 50 年代末的斯坦福大学和普渡大学等先驱机构，当时的名称不同，通常是数学系或电气工程系的一个分支。它建立了一个基于廉价鼓式计算机的滩头阵地，包括 Librascope LGP30，尤其是 IBM 650。20 世纪 50 年代中期，几所顶尖大学的暑期学校课程进一步使该学科合法化^。57

These examples show that there was more to software than the development and evolution of programming languages. But programming languages came to dominate the academic discipline that arose during this period, namely, computer science. The discipline first appeared in the late 1950s at pioneering institutions, including Stanford and Purdue, under different names and often as a division of the Mathematics or Electrical Engineering Departments. It established a beachhead based on inexpensive drum-based computers, including the Librascope LGP30 and especially the IBM 650. Summer school sessions at several top universities in the mid-1950s further legitimized the discipline.⁵⁷

20 世纪 60 年代初，计算机科学一直在努力定义自己和它的目的，不仅与电气工程和应用数学等既定学科有关，而且与校园内使用计算机进行会计、记录和行政工作有关（并与之不同）。58 在^{负责该学科的人中，斯坦福大学数学系的乔治·福赛斯教授可能是最具影响力的}^。在他的推动下，数学系于 1961 年开设了一个计算机科学系；1965 年，斯坦福大学成立了一个独立的系，这是该国最早的系之一，至今仍是最受推崇的系之一。59

In the early 1960s computer science struggled to define itself and its purpose, in relation not only to established disciplines of electrical engineering and applied mathematics, but also in relation to—and as something distinct from—the use of computers on campus to do accounting, record keeping, and administrative work.⁵⁸ Among those responsible for the discipline that emerged, Professor George Forsythe of Stanford’s mathematics faculty was probably the most influential. With his prodding, a Division of Computer Science opened in the mathematics department in 1961; in 1965 Stanford established a separate department, one of the first in the country and still one of the most well-regarded.⁵⁹

1967 年秋，匹兹堡卡内基理工学院的赫伯特·西蒙、艾伦·佩利斯和艾伦·纽厄尔给《科学》杂志的编辑写了一封言辞流利、影响深远的信，信中他们将计算机科学定义为“对计算机的研究”。60^这个定义隐含着这样一种观念：计算机不是静态的产物，而是一个根据一套规则执行动态过程的系统。他们为这个定义和计算机科学的合法性辩护，驳斥了许多反对意见，其中最明显的一个就是计算机是人造现象，因此对计算机的研究不能算作自然科学。西蒙（他因在管理科学领域的工作而获得诺贝尔奖）认为，许多自然科学研究的现象并非完全脱离人类创造，而且无论如何，将对人工智能的研究变成一门科学并没有错。第二年，他发表了一系列演讲，进一步发展了这一论点，这些演讲出版为《人工智能科学^》。61

In the fall of 1967 Herbert Simon, Alan Perlis, and Allen Newell, all of the Carnegie Institute of Technology in Pittsburgh, wrote an eloquent and influential letter to the editor of Science, in which they defined computer science as “the study of computers.”⁶⁰ Implicit in that definition was the notion of a computer not as a static artifact but as a system that carried out dynamic processes according to a set of rules. They defended that definition and the legitimacy of computer science against a number of objections, including the obvious one that computers were a man-made phenomenon and hence their study could not be one of the natural sciences. Simon (who won a Nobel Prize for his work in what might be called management science) argued that many natural sciences studied phenomena that were not totally divorced from human creation, and anyway, there was nothing wrong with making a science of the study of the artificial. The following year he delivered a series of lectures that further developed this argument, published as The Sciences of the Artificial.⁶¹

这三人提出并驳斥的一个反对意见是，计算机科学实际上是“算法（或程序）的研究，而不是计算机的研究”。62^他们认为这样的定义过于狭隘。他们的反驳很无力——他们给出的理由之一是他们的专业协会被称为计算机协会。无论如何，计算机科学在随后的几年里发展到他们所说的不是的确切含义——算法研究，重点是更狭窄的编程语言领域。尽管有 ACM 这个名字，但电气工程仍然存在硬件问题。在《科学》杂志上发表这封信六个月后，ACM 发布了课程 '68，该协会认为这套课程将为本科生提供在智力上站得住脚的计算机科学基础。63^课程'68 强调计算机科学的数学和理论基础。没有计算机硬件的研究。1965 年发布的早期版本推荐（作为选修课）电子和模拟计算机课程；这些课程从 1968 年的课程中删除。如果学生想学习“计算机”（Newell、Simon 和 Perlis 定义计算机），那么他们还必须学习电气工程。ACM 选择强调算法程序、编程语言和数据结构。一位评论家称，1968 年课程对计算机科学学科的影响不亚于 1945 年的 EDVAC 报告。64到¹⁹⁶⁸年，计算机科学已获得尊重。本科生可以在 100 所美国大学中获得计算机科学学位，而 1964 年只有 12 所。到 1980 年代，它已成为美国几乎每个校园最受欢迎的本科专业之一^。65

One objection that the trio raised and then refuted was that computer science was really “the study of algorithms (or programs), not computers.”⁶² They felt that such a definition was too restrictive. Their refutation was weak—among the reasons they gave was that their professional society was called the Association for Computing Machinery. In any event, computer science evolved in subsequent years to mean precisely what they said it was not—the study of algorithms, with a focus on the even narrower field of programming languages. The ACM’s name notwithstanding, hardware issues remained with electrical engineering. Six months after the appearance of the letter in Science, the ACM published Curriculum ’68, a set of courses that the association felt would provide an intellectually defensible grounding in computer science for undergraduates.⁶³ Curriculum ’68 emphasized a mathematical and theoretical basis for computer science. The study of computer hardware was absent. An earlier version published in 1965 recommended (as electives) courses in electronics and analog computers; these were dropped from the 1968 curriculum. A student wanting to study “computers,” as Newell, Simon, and Perlis defined them, would have to study Electrical Engineering as well. The ACM chose to emphasize algorithmic procedures, programming languages, and data structures. One critic called Curriculum ’68 as influential on the discipline of computer science as the 1945 EDVAC Report.⁶⁴ By 1968 computer science had gained respectability. An undergraduate could obtain a degree in it at one of one hundred U.S. universities, up from only twelve in 1964. By the 1980s it had become one of the most popular undergraduate majors on nearly every campus in the United States.⁶⁵

1968 年和 1969 年的其他事件

Other Events of 1968 and 1969

1968 年和 1969 年的一系列类似事件进一步确立了软件的地位、它与计算机科学的关系以及它与工业、商业和军事计算机用户的关系。

In 1968 and 1969 a cluster of similar events further established the place of software, its relationship to computer science, and its relationship to industrial, commercial, and military computer users.

唐纳德·E·克努斯

Donald E. Knuth

1968 年，唐纳德·E·克努斯出版了计划共七卷的《计算机编程艺术》系列丛书的第一卷。用他的话来说，《基本算法》以印刷形式记录了那些作为“民间传说……但理论开发相对较少”的编程技术。66^其他人也曾尝试将计算机编程建立在坚实的理论基础之上，但这些尝试往往并不能为解决实际的编程问题提供太多实际帮助。大多数编程教学都与特定机器的特性密切相关，包括某些寄存器的二进制代码、往返于磁盘或磁鼓存储器的信号时序等等。克努斯为实用的计算提供了理论基础，他的书将其确立为算法，即一种正式的程序，在实际计算机的限制下，人们可以使用它在合理的时间内解决问题。

In 1968 Donald E. Knuth published the first of a projected seven volumes of a series on The Art of Computer Programming. Fundamental Algorithms, in his words, set down in print techniques of programming that had been passed along as “folklore, … but [for which] comparatively little theory had been developed.”⁶⁶ Others had also attempted to place computer programming on solid theoretical ground, but often these attempts did not offer much practical help in solving actual programing problems. Most teaching of programming was intimately bound up with the idiosyncrasies of a particular machine, including the binary codes for certain registers, the timing of signals to and from disk or drum stores, and so on. Knuth provided a theoretical basis for computing that was practical, and his books established it as the algorithm, a formal procedure that one can use to solve a problem in a reasonable length of time, given the constraints of actual computing machines.

结构化编程

Structured Programming

1968 年 3 月，在发表《课程 '68》的同一期 ACM 期刊上，刊登了一封致编辑的信，标题很奇怪，叫《Go-To 声明被认为有害》。67^这封信的开头是这样的：

In March 1968, in the same issue of the ACM journal that published Curriculum ’68, there appeared a letter to the editor with the curious title “Go-To Statement Considered Harmful.”⁶⁷ The letter opened with the statement,

多年来我一直熟知这样的观察：程序员的素质是他们所编写程序中 go to 语句频率的递减函数。

For a number of years I have been familiar with the observation that the quality of programmers is a decreasing function of the frequency of go to statements in the programs they produce.

这位作者是荷兰埃因霍温理工大学的 Edsger Dijkstra。他的信引发了一场持续数年的争论。这只是他长期努力推动计算机科学走向更正式的理论基础的一小部分，但这封信的大胆主张成为了他试图做的更复杂工作的象征，他将复杂的软件系统从沙地基础转移到基础理论的基石上。

The author was Edsger Dijkstra of the Technical University of Eindhoven, the Netherlands. His letter set off an argument that continued for the next few years. It was only a minor part of his long effort to move computer science toward a more formal theoretical basis, but the letter’s bold assertion became a symbol for the more complex work he was trying to do to move complex software systems from foundations of sand to the bedrock of basic theory.

随后，围绕 Dijkstra 关于“go to”的断言，人们展开了一场漫长而激烈的争论。几乎没有参与者意识到 Dijkstra 关心的不仅仅是某个特定的命令。工业界的批评者认为，这场争论是学术计算机科学无关紧要的又一个例子。从短期来看，这封信推动了“结构化编程”概念的发展，其追随者认为这种方法将使编程从一门艺术（Knuth 在他的书名中这样称呼它）发展成为一门科学。无论 Dijkstra 的信是否是推动力，编程在接下来的几年里确实朝着这个方向发展。

There followed a long and acrimonious debate over Dijkstra’s assertion about “go to.” Few participants seemed to realize that Dijkstra was concerned with something more profound than one particular command. Critics in the industrial world saw the fuss as one more example that proved the irrelevance of academic computer science. In the short term, the letter gave a push to the concept of “structured programming,” a method that its adherents believed would advance programming from an art (as Knuth called it in the title of his books) to a science. Whether Dijkstra’s letter was the impetus or not, programming did move in this direction in the following years.

知识产权问题

Intellectual Property Issues

接下来与ACM通讯编辑的信件往来涉及另一个问题，这个问题也对编程质量产生了深远的影响^。68这次信件往来于密歇根大学的 Bernard Galler 教授（MAD 的创建者之一）和 Calvin Mooers 之间，Mooers 开发了一种他称之为“TRAC 语言”的语言。Mooers 是 Rockford 研究所的创始人之一，该研究所曾寻求对 TRAC 语言的法律保护，这种保护将阻止任何人更改、修改或扩展它。Galler 指出，最好和最成功的语言是那些受益于用户（包括研究生）的语言，他们在使用过程中对其进行了改进或修改。双方各有道理。如果没有某种保护，一种语言的一个版本，尽管“改进了”，也会与另一种版本不兼容，仅这一点就可能使任何可能声称的改进失效。但正如 SHARE 和 MAD 的例子所证明的那样，软件开发需要用户来进步。部分由于 Mooers 的政策，他的语言几乎没有用处^。69

The very next exchange of letters to the editor of the Communications of the ACM concerned another issue, that also came to have a profound effect on the quality of programming.⁶⁸ The exchange was between Professor Bernard Galler of the University of Michigan (one of the creators of MAD), and Calvin Mooers, a developer of a language he called “the TRAC language.” The Rockford Research Institute, of which Mooers was a founder, had sought legal protection for the TRAC language, protection which would have prevented anyone from altering, modifying, or extending it. Galler pointed out that the best and most successful languages were those that had benefitted from users, including graduate students, who had improved or modified them in use. Both sides had a point. Without some protection, one version of a language, albeit “improved,” would be incompatible with another, and this alone could vitiate whatever improvements might be claimed. But as the examples of SHARE and MAD proved, software development needed the user to progress. In part because of Mooers’s policy, his language found little use.⁶⁹

八年后，在个人电脑的早期发展中，商业软件的所有权和控制权问题变得至关重要。随着微软公司澄清了这个问题并确立了自己的权利，微软不仅成为了一家占主导地位的软件供应商，还为众多供应商生产大量廉价软件奠定了基础。70 ^（这将在第 7 章中进一步讨论。）

Eight years later, during the early development of the personal computer, the question of ownership and control of commercial software became crucial. As the Microsoft Corporation clarified that question and established its rights, not only did Microsoft become a dominant software provider, it also set the stage for the production of large amounts of inexpensive software from a myriad of vendors.⁷⁰ (This will be discussed further in chapter 7.)

软件工程

Software Engineering

1968 年 10 月，德国加米施召开了一次会议，会议的主题是“软件工程”。71^这次会议标志着天真时代的结束，人们意识到软件生产中的“危机”不会很快结束，但可以而且必须加以管理。这项活动的名称——工程——是故意带有挑衅性的，表明软件危机背后的原因是程序员缺乏传统工程领域中的理论基础和日常实践纪律。这次会议由北约赞助，这进一步表明软件工程与以大学为中心的计算机科学的距离。72^会议组织者已经认识到，计算机负责危及人类生命的系统，包括北约使用的军事系统。这些系统不能容忍在批量数据处理环境中可能只是小麻烦的“错误”。这表明它与其他形式的工程相似，尤其是土木工程，人们通常信任一座桥梁，因为他们信任设计它的人。然而，在其他方面，这种类比就不成立了。在土木工程领域，多年来形成了认证传统和法律责任链；而在计算机领域，这种传统却没有建立，也不会出现。试图控制谁可以自称“计算机程序员”的尝试似乎总是徒劳无功。部分原因是，新技术（如个人工作站）为新进入者提供了进入该领域的途径，而该行业的资深人士无法控制这一点。1996 年在德国举行的软件工程史会议得出了一个意想不到的结论：建立软件工程的尝试整体上失败了^。73

In October 1968 a conference was convened in Garmisch, Germany, with the provocative title “Software Engineering.”⁷¹ The conference marked the end of the age of innocence, a realization that a “crisis” in software production would not end soon but could—and had to—be managed. The name given to that activity—engineering—was deliberately provocative, suggesting that what lay behind the software crisis was the fact that programmers lacked the theoretical foundations and disciplines of daily practice that one found in traditional fields of engineering. That the conference was sponsored by NATO further revealed software engineering’s distance from computer science, which was centered in the universities.⁷² Conference organizers had recognized that computers were responsible for systems that put human lives at risk, including the military systems employed at NATO. These systems could not tolerate a “bug” that in a batch data processing environment might be only a minor irritant. That suggested a parallel with other forms of engineering, especially civil engineering, where people routinely trusted a bridge because they trusted the people who designed it. In other respects, however, the analogy broke down. In civil engineering, a tradition of certification and a chain of legal responsibility had evolved over the years; no such tradition had been established in computing and none would emerge. Attempts to control who might claim the title “computer programmer” seem always to be futile. Part of the reason is that new technology, like the personal workstation, offers an avenue for new entrants into the field, and established members of the profession cannot control that. A 1996 conference on the history of software engineering, also held in Germany, came to the unintended conclusion that the attempt to establish software engineering on the whole had failed.⁷³

拆分

Unbundling

1968 年 12 月，在美国政府的压力下，IBM 宣布次年将“拆分”其软件；即单独收取软件费用，而不是将其成本与硬件系统成本合并。74^它开始销售的第一批产品之一也是计算机历史上最成功的产品之一，即客户信息控制系统 (CICS)，它从 1969 年 7 月开始以磁带形式提供，每月收费 600 美元。75^软件仍然虚无缥缈，但现在可以买卖了。这一决定的影响是向商业供应商开放了软件领域，他们现在将受到自由市场强大而无情的力量的驱动。

In December 1968, under pressure from the U.S. government, IBM announced that the following year it would “unbundle” its software; that is, charge separately for it instead of combining its costs with that of the hardware systems.⁷⁴ One of the first products it began to sell was also one of the most successful in the history of computing, its Customer Information Control System (CICS), which it offered on a tape beginning in July 1969 for $600 a month.⁷⁵ Software remained ethereal, but now it could be bought and sold. The effect of that decision was to open up the field of software to commercial vendors, who would now be driven by the powerful and unforgiving forces of the free marketplace.

学术界的动向最终影响到了软件公司。尽管软件危机一直存在，但编程变得更结构化，更牢固地基于理论。与此同时，计算机行业正经历最具创新性的时期之一。20 世纪 60 年代末，集成电路开始出现在商业系统中，这立即催生出了一种内存要求有限的新型廉价计算机。这些计算机没有空间来实现高度结构化的语言，如 FORTRAN 的批评者所偏爱的 Pascal。因此，程序员只能求助于非结构化的机器语言或汇编语言，等等。许多人希望将软件置于更正式的基础上，但这迟迟没有实现。

The effects of what was going on in the academic world eventually made their way to the software houses. Programming became more structured and more firmly based on theory, although the software crisis became a permanent fixture. Meanwhile, the computer industry was going through one of its most innovative periods. It was in the late 1960s that the integrated circuit began to show up in commercial systems, which immediately created a new class of inexpensive computers that had limited memory requirements. These computers had no room to implement the highly structured languages, like Pascal, that critics of FORTRAN preferred. So programmers resorted to unstructured machine or assembly language, “go to” and all. Putting software on a more formal basis, which so many had hoped for, would arrive late, if ever.

1969 年，新泽西州贝尔电话实验室的 Ken Thompson 和 Dennis Ritchie 开始研究 UNIX 操作系统。他们使用的计算机是数字设备公司的 PDP-7，即使以当时的标准来看，这台机器的内存也极其有限。76 ^Thompson用汇编语言编写了该系统的早期版本，但很快他和同事就开发了一种名为“B”的“系统编程语言”，到 1973 年，这种语言演变为 C 语言^。77用 Ritchie 的话来说，C 是一种“接近机器”的语言，这一特点反映了其创建者希望保留汇编语言的强大功能。78^从这个意义上讲，它违背了当时所推崇的结构化编程原则。这一特性也使其成为个人计算机时代最受欢迎的语言之一，微软等 20 世纪 90 年代的大型软件公司都采用这种语言。

In 1969 Ken Thompson and Dennis Ritchie at the Bell Telephone Laboratories in New Jersey began work on what would become the UNIX operating system. The computer they used was a Digital Equipment Corporation PDP-7, a machine with extremely limited memory even by the standards of that day.⁷⁶ Thompson wrote the earliest version of this system in assembler, but soon he and his colleagues developed a “system programming language” called “B,” which by 1973 had evolved into a language called C.⁷⁷ C is a language that is “close to the machine,” in Ritchie’s words, a characteristic that reflected its creators’ desire to retain the power of assembly language.⁷⁸ To that extent it went against the tenets of structured programming that were then being espoused. That quality also made it one of the most popular languages of the personal computer era, as practiced at the giant software houses of the 1990s such as Microsoft.

UNIX 和 C 的成功抵消了 20 世纪 60 年代后期两种失败的编程语言 ALGOL-68 和 PL/I 的影响。这些失败的语言一方面表明了学术计算机科学与用户之间最严重的鸿沟，另一方面表明了人们早已熟知的认识：委员会并不适合进行良好的软件开发。

The success of UNIX and C balanced two failed programming languages of the late 1960s, ALGOL-68 and PL/I. These failed languages illustrated, on the one hand, the worst of the gulf between academic computer science and the users, and on the other hand, the already familiar realization that committees are ill-suited to good software development.

ALGOL-68 是 ALGOL-60 的改进版本，在学术界之外，只有 Burroughs 公司支持 ALGOL-60。当时，ALGOL-60 失败的原因引起了广泛讨论。许多人认为，IBM 对 FORTRAN 的支持注定了它的失败，但我们将会看到，IBM 对 PL/I 的支持无力挽救它。更严重的是 ALGOL-60 缺乏 I/O 规范。对于仍在争论是否要使用高级语言的计算机用户来说，这是一个致命的疏忽。79 ²⁰世纪 60 年代中期，国际信息处理联合会 (IFIP) 成立了一个工作组来扩展 ALGOL 语言，并发布了新版本，即 ALGOL-68。80 它^也因不同的原因在市场上失败了。 ALGOL-60 基于形式化结构，非常精简，而 ALGOL-68 则因为试图做太多事情而负担过重，结果导致某些功能干扰了其他功能的干净实现。这很难理解。为了满足全球广大用户的需求，委员会做出了一些让少数人满意的东西。81^它在少数地方实施，特别是在英国的 ICL 计算机上。否则，它的主要遗产可能是 Pascal 语言，这是瑞士联邦技术学院的 Nicholas Wirth 为应对 ALGOL-68 的复杂性而开发的一种结构严密的语言^。82

ALGOL-68 was an attempt to improve upon ALGOL-60, which outside the academy only the Burroughs Corporation supported. The reasons for ALGOL-60’s failure was the subject of much discussion at the time. Many believed that IBM’s support for FORTRAN doomed it, but we shall see that IBM’s support for PL/I did not have the power to save it. More serious was ALGOL-60’s lack of I/O specifications. For computer users who were still debating whether to use any high-level language at all, that was a fatal omission.⁷⁹ In the mid-1960s, the International Federation for Information Processing (IFIP) established a working group to extend the ALGOL language, and they released a new version of it, which came to be known as ALGOL-68.⁸⁰ It, too, failed in the marketplace, but for different reasons. Whereas ALGOL-60 was based on a formal structure and was very lean, ALGOL-68 was burdened by an attempt to do too much, with the effects that some features interfered with the clean implementation of others. It was hard to understand. In an attempt to satisfy a broad range of users worldwide, the committee produced something that satisfied few.⁸¹ It was implemented in a few places, notably on ICL computers in the U.K. Otherwise, its chief legacy may have been the language Pascal, the tightly structured language that Nicholas Wirth of the Swiss Federal Technical Institute developed in reaction to ALGOL-68’s complexity.⁸²

IBM 也曾努力开发一种语言来取代古老的 FORTRAN 和 COBOL。1964 年，IBM 宣布推出 System/360 系列计算机来替代其商业和科学计算机（参见第 5 章），该公司认为同样可以为这两种应用程序开发一种新语言。1963 年初，IBM-SHARE 联合委员会得出结论，新语言不会是 FORTRAN 的扩展，尽管该语言的现有版本 FORTRAN-IV 非常流行且使用广泛。新语言 PL/I（编程语言一）借鉴了 FORTRAN、COBOL 和 ALGOL。初步版本于 1964 年发布，但当完整语言准备就绪时，COBOL 和 FORTRAN-IV 已在 System/360 系列上站稳脚跟，无法被取代。PL/I 的复杂性掩盖了它的许多优势，包括 IBM 支持它以及它适用于商业和科学的优势。许多 IBM 安装都提供 PL/I，但它从未流行起来。

At IBM there was a similar effort to develop a language to replace the venerable FORTRAN and COBOL. When in 1964 IBM announced its System/360 series of computers as a replacement for both its business and scientific computers (see chapter 5), the company assumed that a new language could likewise be developed for both applications. A joint IBM-SHARE committee concluded in early 1963 that the new language would not be an extension of FORTRAN, even though the existing version of that language, FORTRAN-IV, was very popular and heavily used. The new language, PL/I (Programming Language, One), drew from FORTRAN, COBOL, and ALGOL. Preliminary versions were released in 1964, but by the time the full language was ready, COBOL and FORTRAN-IV had established a foothold on the System/360 series and could not be dislodged. PL/I’s complexity overwhelmed its many advantages, including the advantage that IBM was supporting it and that it was suitable for both business and science. Many IBM installations made PL/I available, but it never became very popular.

结论

Conclusion

计算机编程这项活动是计算机先驱们未曾预见到的。在 20 世纪 50 年代，他们和他们的客户逐渐意识到：首先，它存在；其次，它很重要；第三，值得努力构建工具来帮助完成它。这些工具与应用程序相结合，统称为“软件”，这个术语于 1959 年左右开始使用。83 ^Grace Hopper 和 Maurice Wilkes 等人最初专注于构建子程序库，然后让计算机调用这些子程序并将它们链接在一起以解决特定问题。这让位于更通用的高级计算机语言概念，计算机根据对程序员指定内容的仔细分析生成新的机器代码，这类似于代数和英语的结合。

The activity known as computer programming was not foreseen by the pioneers of computing. During the 1950s they and their customers slowly realized: first, that it existed; second, that it was important; and third, that it was worth the effort to build tools to help do it. These tools, combined with the applications programs, became collectively known as “software,” a term that first came into use around 1959.⁸³ People like Grace Hopper and Maurice Wilkes initially focused on building up libraries of subroutines, and then getting the computer to call these up and link them together to solve a specific problem. That gave way to a more general notion of a high-level computer language, with the computer generating fresh machine code based on a careful analysis of what the programmer specified, in something that resembled a combination of algebra and English.

尽管软件取得了长足进步，但编程似乎总是处于危机状态，并且似乎总是在追赶硬件的进步。这场危机在 1968 年达到顶峰，当时集成电路和磁盘存储正在对硬件系统产生影响。那一年，学术界和贸易界明确承认了这场危机，并成为北约赞助的一次会议的主题，呼吁人们进一步关注这场危机。提出的一些解决方案包括软件工程的新学科、更正式的结构化编程技术以及将取代古老但过时的 COBOL 和 FORTRAN 的新编程语言。尽管 IBM 决定将其软件和服务与硬件分开销售，但这一决定可能对解决问题更有帮助，尽管这一决定并非为了应对这场危机。它导致了商业软件行业需要生产可靠的软件才能生存。然而，危机仍然存在，并成为计算的一个永久方面。软件在 1968 年成熟；接下来的几十年将看到进一步的变化和对硬件进步的进一步适应。

Despite great strides in software, programming always seemed to be in a state of crisis and always seemed to play catch-up to the advances in hardware. This crisis came to a head in 1968, just as the integrated circuit and disk storage were making their impact on hardware systems. That year, the crisis was explicitly acknowledged in the academic and trade literature and was the subject of a NATO-sponsored conference that called further attention to it. Some of the solutions proposed were a new discipline of software engineering, more formal techniques of structured programming, and new programming languages that would replace the venerable but obsolete COBOL and FORTRAN. Although not made in response to this crisis, the decision by IBM to sell its software and services separately from its hardware probably did even more to address the problem. It led to a commercial software industry that needed to produce reliable software in order to survive. The crisis remained, however, and became a permanent aspect of computing. Software came of age in 1968; the following decades would see further changes and further adaptations to hardware advances.

从大型机到小型计算机，1959-1969 年

From Mainframe to Minicomputer, 1959–1969

这个房间可能是好莱坞反乌托邦电影《银翼杀手》或《巴西》的制片人设计的。目之所及之处，都是 IBM Model 027 打孔机，这种机器可以在 80 列卡片上打出矩形孔，每张卡片的标准尺寸为 3-1/4 × 7-3/8 英寸。每个工位都坐着一名女性，她的头向左倾斜，扫描一张装在框架上的纸张，右手灵巧地在机器的按键上移动。每次按下按键，机器都会在卡片上打一个孔，并发出“咚”的一声。一名操作员说，当房间全速运转时，“有一种节奏、节拍和声音”，让每位操作员都知道她正在完成工作。1 一位^数据处理经理的看法略有不同：他说声音“就像你戴着头盔，有人用锤子敲打它。” ²有人拍了一部关于这次操作的电影；从配乐中，人们可能会得出这样的结论：如果说有什么不同的话，那就是第二种观点是保守的。

The room could have been designed by Hollywood producers of dystopian films like Blade Runner or Brazil. As far as the eye could see were rows of IBM Model 027 keypunches—machines that punched rectangular holes into 80-column cards, each of them a standard 3-1/4 × 7-3/8 inches. Seated at each station was a woman, her head tilted to the left to scan a piece of paper mounted on a frame, her right hand deftly floating over the keys of the machine. Each press of a key caused it to punch a hole in a card with a solid “thunk.” When the room was in full swing, said one operator, “there was a certain rhythm, a beat, a sound” that let each operator know she was getting the job done.¹ A data processing manager had a slightly different opinion: he said the sound was “like you had a helmet on and someone was hitting it with a hammer.”² A film was made of the operation; from its soundtrack one might conclude that, if anything, the second opinion was conservative.

这个房间是美国国税局在 20 世纪 60 年代中期设立的几个处理纳税申报的区域中心之一。当时，国税局已经采用了电子数字计算机，拥有世界上最先进、最复杂的系统之一。其核心是位于西弗吉尼亚州马丁斯堡的国家中心的一套 IBM 大型机。1964 年，也就是电影拍摄的时候，该中心正在使用一套晶体管 IBM 7070，这是第2 章中讨论的 7090 的商业^版本。3

The room was in one of several regional centers set up in the mid-1960s by the U.S. Internal Revenue Service to process tax returns. By then the IRS had embraced the electronic digital computer and owned one of the most sophisticated and complex systems in the world. At its heart was a set of IBM mainframes at a national center in Martinsburg, West Virginia. In 1964, around the time the film was made, the Center was using a set of transistorized IBM 7070s, business versions of the 7090 discussed in chapter 2.³

就在这些妇女们打字打出纳税申报单的同时，伍斯特（马萨诸塞州）的《电报和公报》也进入了计算机时代。一些大城市的报纸已经安装了大型计算机来取代使用热铅排版的 Mergenthaler Linotype 机。伍斯特报纸通过从马萨诸塞州梅纳德的数字设备公司购买了一台小得多但功能强大的 30,000 美元“计算机排版系统”加入了这一运动。这套系统比几张办公桌大不了多少。它的核心是 Digital 提供的一种新型计算机，即 PDP- ^8。4

At the same time that the women were keypunching tax returns, the Worcester (Massachusetts) Telegram and Gazette was also entering the computer age. A few big-city newspapers had already installed mainframes to replace the Mergenthaler Linotypes that set type in hot lead. The Worcester paper was able to join this movement by purchasing a much smaller, but very capable, $30,000 “Computer Typesetting System” from the Digital Equipment Corporation of Maynard, Massachusetts. The system was not much bigger than a couple of office desks. At its heart was a new type of computer offered by Digital, a PDP-8.⁴

数字计算机在 20 世纪 60 年代初期初露头角，但最终在许多商业、会计和政府业务中占据了稳固的地位。现在，它们还出现在许多新的应用中。

Digital computers began the decade of the 1960s with a tentative foothold; they ended with an entrenched position in many business, accounting, and government operations. They were also now found in a host of new applications.

推动这一运动的力量既有技术力量，也有社会力量。前者包括晶体管取代真空管，以及 FORTRAN 和 COBOL 等语言的开发，这些语言使编程变得更加容易。后者包括联邦政府对记录保存的需求增加，这是“大社会”计划带来的。1961 年 5 月，约翰·肯尼迪总统提出了在 20 世纪 60 年代末将人类送上月球并安全返回的挑战，这使美国太空计划变成了一个研究和生产中心的综合体，预算无限，对计算能力的需求无穷无尽。美国正进入经济增长和繁荣的十年，同时在州际公路、郊区住房和喷气式飞机方面进行了大规模投资。所有这些都给基于打孔卡制表机、机械加法机和计算器的信息处理程序带来了压力。

The forces driving this movement were both technical and social. Among the former were the introduction of transistors in place of vacuum tubes and the development of languages like FORTRAN and COBOL that made programming easier. Among the latter was the increased demand for record-keeping by the federal government, brought on by programs of the “Great Society.” President John Kennedy’s challenge, in May 1961, to put a man on the moon and return him safely by the end of the decade transformed the U.S. space program into a complex of research and production centers with unlimited budgets and insatiable appetites for computing power. The United States was entering a decade of economic growth and prosperity, accompanied by major investments in interstate highways, suburban housing, and jet aircraft. All of these put a strain on information-processing procedures that were based on punched card tabulators, mechanical adding machines, and calculators.

20 世纪 60 年代的计算机行业不仅仅是现有公司向现有客户销售大量新机器的故事，尽管这种情况确实发生过。那十年也见证了该行业的性质发生了变化。从某种意义上说，计算机又一次被重新发明了。就像埃克特和莫奇利将快速计算器 ENIAC 改造成通用数据处理和科学设备一样，现在像数字设备公司这样的新公司也重新设计了计算机的内部架构、编程、营销方式和应用。

Computing in the 1960s was not just a story of existing firms selling new machines in increasing volume to existing customers, although that did occur. That decade also saw the nature of the industry transformed. In a sense the computer was reinvented yet again. Just as Eckert and Mauchly transformed a fast calculator, the ENIAC, into a general-purpose data processing and scientific device, so now did new companies like the Digital Equipment Corporation rework the computer’s internal architecture, its programming, the way it was marketed, and the applications it was used for.

计算机技术进步的速度和现有产品的快速淘汰速度在其他行业中是少有的。IBM 非常了解这一事实，自 20 世纪 50 年代末以来，其市场份额一直徘徊在 70% 左右。5 ^IBM建立了一个大型研究部门，在美国东西海岸以及欧洲设有主要实验室，以确保能够跟上固态电子、磁带和磁盘存储、编程语言和逻辑电路的发展。其中一些研究是在固态物理和数学的基础层面进行的，这几乎不可能快速产生回报，但计算的性质决定了必须进行此类研究。应用研究可能会产生更快的真空管或更好的卡片打孔机（这些设备确实被发明了），但它不会产生大众媒体认为具有革命性的计算进步^。6

The rate of technological advance in computing, and the rapid obsolescence of existing products, had few counterparts in other industries. It was a fact that was well understood by IBM, whose market share hovered around 70 percent from the late 1950s onward.⁵ IBM built up a large research department, with major laboratories on both coasts as well as in Europe, to ensure that it could stay abreast of developments in solid-state electronics, tape and disk storage, programming languages, and logic circuits. Some of that research was conducted at fundamental levels of solid-state physics and mathematics, which offered little chance of a quick payoff, but the nature of computing dictated that such research be done. Applied research might have produced a faster vacuum tube or a better card punch (and such devices were indeed invented), but it would not produce the kinds of advances in computing that are regarded in the popular press as revolutionary.⁶

支持此类研究虽然花费不菲，但却让 IBM 比其竞争对手更具优势，因为竞争对手的客户群更小。而且，它为任何想进入该行业的新公司设置了很高的门槛。如果新进入者想利用 IBM 开发的一项全新技术，就必须构建和销售一个平衡的系统，包括软件 — IBM 可能更擅长生产所有这些软件。前面提到的 Philco 就是这样一家公司。Philco 开发的表面势垒晶体管使其领先 IBM 一年，而 IBM 在 20 世纪 50 年代末刚刚推出真空管 709。7^但是，当 IBM 推出晶体管 7090 时，Philco 无法跟上竞争的步伐，并于 1964 年退出了该业务^。8为了生存，新进入该领域的公司除了拥有卓越的技术外，还必须拥有 IBM 和其他大型机公司服务不佳的利基市场。 IBM 的主导地位意味着它能够以适当的速度进行技术进步，而不会使其现有客户群过快过时。20 世纪 50 年代末，几乎没有制造商（IBM 除外）通过销售大型计算机系统获利。

Supporting such research, though expensive, gave IBM an advantage over its competitors, who had to work from a smaller customer base. And it created a high barrier to any new firm wishing to enter the industry. If a newcomer wanted to exploit a radically new piece of technology IBM had developed, it would have to build and market a balanced system, including software—all the pieces of which IBM was probably also better at producing. One such company was Philco, discussed earlier. The surface-barrier transistors that Philco developed put it a year ahead of IBM, which in the late 1950s had just introduced the vacuum-tube 709.⁷ But once IBM countered with its transistorized 7090, Philco could not maintain the pace of competition and left the business by 1964.⁸ In order to survive, a new entrant into the field had to have, in addition to superior technology, a niche that was poorly served by IBM and the other mainframe companies. IBM’s dominant position meant that it could mete out technical advances at a pace that did not render its installed base obsolete too quickly. Almost no manufacturers save IBM made a profit selling large computer systems in the late 1950s.

尽管批评者们指责 IBM，但 IBM 并不总是能够控制创新的步伐。例如，它放弃真空管 709 的速度比它想要的要快；十年后，它以令人不安的速度从 System/360 过渡到 System/370。IBM 确实成功的一点是保持了穿孔卡的基本输入介质的可行性，以及通过客户安装的基本数据流。同一张卡片，以相同的方式编码，使用自 20 世纪 30 年代以来几乎没有变化的键盘穿孔机，在 20 世纪 60 年代及以后一直为 IBM 的计算机服务。穿孔卡操作中隐含的顺序处理和文件结构也以 20 世纪 60 年代大多数大型计算机中心常见的批处理形式幸存下来。这减轻了许多客户采用新技术的冲击，并确保了 IBM 在这些站点的计算中持续发挥影响力。

Despite what its critics charged, IBM did not always succeed in controlling the pace of innovation. It abandoned the vacuum-tube 709 faster than it wanted to, for example; ten years later it made an uncomfortably quick transition from the System/360 to the System/370. One place where IBM did succeed was in keeping viable the basic input medium of the punched card, and with that the basic flow of data through a customer’s installation. The same card, encoded the same way and using a keypunch little changed since the 1930s, served IBM’s computers through the 1960s and beyond. The sequential processing and file structure, implicit in punched card operations, also survived in the form of batch processing common to most mainframe computer centers in the 1960s. That eased the shock of adopting the new technology for many customers, as well as ensuring IBM’s continued influence on computing at those sites.

IBM 由此在行业内创造了一种均衡状态。一些经济学家认为，其 70% 的市场份额“刚好足以”维持创新、稳定和利润，但又不至于导致停滞和其他垄断弊端。9^如果不是外部因素发生根本性变化，计算机行业可能会继续这种状态数十年，就像美国汽车行业由三家公司主导，在 20 世纪 80 年代实现了均衡一样。

IBM thus created a state of equilibrium in the industry. Its 70 percent market share, some economists felt, was “just enough” to maintain innovation, stability, and profits, but not so much as to bring on stagnation and the other evils of monopolization.⁹ Were it not for a radical change in an external factor, the computer industry might have gone on this way for decades, just as the U.S. auto industry, dominated by three firms, achieved an equilibrium into the 1980s.

联邦政府的影响

The Influence of the Federal Government

外部因素是美国国防部，其资金对计算机技术的发展至关重要。从 ENIAC 开始，军事支持对计算机来说并不是什么新鲜事。改变的是国防支持下的研究性质，特别是在 1950 年朝鲜战争爆发后。1950 年后，军事对物理、电气工程和数学基础研究的支持急剧增加。这些研究的性质也发生了变化，从军方详细指定其需求，到研究人员自己（主要大学的教授和他们的研究生）积极参与定义工作性质和目标。10^军事资金被引导到著名大学的研究部门，为大型工业实验室提供了另一种知识来源。这些知识反过来又使知名公司以外的个人能够进入计算机行业。它对计算的影响是巨大的。

The external factor was the U.S. Defense Department, whose funding was crucial to the advance of computer technology. Military support was nothing new to computing, beginning with the ENIAC. What changed was the nature of research done under defense support, especially after the onset of the war in Korea in 1950. Military support for basic research in physics, electrical engineering, and mathematics increased dramatically after 1950. The nature of that research also changed, from one where the military specified its needs in detail, to one where the researchers themselves—professors and their graduate students at major universities—took an active role in defining the nature and goals of the work.¹⁰ Military funding, channeled into research departments at prestigious universities, provided an alternative source of knowledge to that generated in large industrial laboratories. This knowledge, in turn, allowed individuals outside established corporations to enter the computer industry. Its effect on computing was dramatic.

本章首先通过案例研究来说明联邦政府其他部门作为客户如何影响计算；也就是说，他们对计算的大量需求加上慷慨的预算如何刺激大型主机安装的增长，其中 IBM 7000 系列计算机占主导地位。随后，本章从另一个角度看待计算，并描述固态物理和电子学研究如何产生一种完全不同类型的计算机，以及一群关键的个人如何利用这项研究来重新定义该行业。

This chapter begins by looking at case studies that illustrate how other branches of the federal government affected computing as customers; that is, how their heavy demands for computation combined with generous budgets spurred the growth of large mainframe installations, dominated by the IBM 7000 series of computers. Following that, the chapter looks at computing from the other side and describes how research in solid-state physics and electronics produced a fundamentally different type of computer, and how a key group of individuals leveraged that research to redefine the the industry.

马萨诸塞州蓝十字

Massachusetts Blue Cross

马萨诸塞州蓝十字是典型的商业客户。1960 年 12 月，经过三年的规划和分析，马萨诸塞州蓝十字/蓝盾购买了一台 IBM 7070 计算机来处理其在制表机上进行的工作。11^{蓝十字打算将 2,500,000 名用户的记录}^放在24 卷磁带上。到 1961 年 6 月，转移工作完成，存放打孔卡和打印记录的 150 个文件柜也退役了。12三年后，7070 升级为 7074；蓝十字还购买了一台较小的 IBM 1401，专门用于输入和输出。尽管到那时 COBOL 和 FORTRAN 都已广泛使用并得到支持，但蓝十字选择使用更原始的语言 AUTOCODER，因为用一位前雇员的话来说，“FORTRAN 和 COBOL……占用了太多主内存，编译处理时间太长。” ¹³（见图4.1。）

Massachusetts Blue Cross was a typical commercial customer. In December 1960, after three years of planning and analysis, Massachusetts Blue Cross/Blue Shield acquired an IBM 7070 computer to process work it had been doing on tabulating machines.¹¹ Blue Cross intended to place, on twenty-four reels of tape, the records of 2,500,000 subscribers. By June 1961 the transfer was completed, and the 150 file cabinets that held the punched cards and printed records were retired.¹² The 7070 was upgraded three years later to a 7074; Blue Cross also acquired a smaller IBM 1401 dedicated solely to input and output. Although by this time COBOL and FORTRAN were both widely available and supported, Blue Cross chose to use instead the more primitive language AUTOCODER, because, in the words of a former employee, “FORTRAN and COBOL …used up too much main memory and took too much processing time for compilation.”¹³ (See figure 4.1.)

1965 年，随着《社会保障法》修正案的通过，医疗保险业务的命运发生了戏剧性的转变，该修正案为 65 岁及以上的美国人建立了医疗保险。马萨诸塞州的蓝十字/蓝盾成功赢得了在该州管理该计划的工作，并于 1966 年秋季成功将账户计算机化。该公司声称是该国第一家完全计算机化医疗保险的公司。然而，蓝十字使用 1961 年革命性的 7070 使处理速度提高了数千倍，但现在却不够了。该公司租用了位于波士顿郊区的另一台 7070 的计算机时间，蓝十字员工每天晚上开着一辆装满纸牌的汽车前往南桥，通宵运行程序，早上开车带着输出结果返回波士顿。

The fortunes of the medical insurance business took a dramatic turn in 1965, with the passage of amendments to the Social Security Act that established Medicare for Americans age sixty-five and over. Blue Cross/ Blue Shield of Massachusetts successfully bid for the job of administering the program in that commonwealth and managed to computerize the account by the fall of 1966. It claimed to be the first in the country to have fully computerized Medicare. However, the thousand-fold increase in processing speeds that Blue Cross got from using a 7070, revolutionary in 1961, was now inadequate. The company rented computer time on another 7070 located in the Boston suburbs, with Blue Cross employees driving a car loaded with decks of cards out to Southbridge every evening, running the programs overnight, and driving back to Boston with the output in the morning.

1967 年，Blue Cross 收购了一台 IBM 新一代 System/360 计算机来处理工作量。到 1960 年代末，现场有三台System/360 以及 7074/1401 系统。43 台磁带驱动器处理记录，计算机操作每周 7 天、每天 24 小时运行。COBOL 现在是首选语言，尽管 AUTOCODE 程序仍在使用，方法是在其中一台 360 上运行模拟器，使其“看起来像”7074/1401 系统。14^该公司继续依赖 IBM 大型机系统。

In 1967 Blue Cross acquired one of IBM’s new-generation System/360 computers to handle the workload. By the end of that decade there were three System/360s on site, as well as the 7074/1401 system. Forty-three tape drives handled the records, and computer operations went on twenty-four hours a day, seven days a week. COBOL was now the preferred language, although the AUTOCODE programs continued to be used by running an emulator on one of the 360s that made it “look like” a 7074/1401 system.¹⁴ The company continued to rely on IBM mainframe systems.

美国宇航局艾姆斯研究中心

NASA-Ames Research Center

美国宇航局艾姆斯研究中心位于加利福尼亚州山景城，自 1940 年成立以来一直是高速空气动力学研究中心，当时隶属于当时的国家航空咨询委员会。1957 年苏联人造卫星的轰动，以及随后肯尼迪总统向国家提出的在 20 世纪 60 年代末之前将人类送上月球的挑战，让该中心产生了一种自二战以来从未有过的紧迫感。该实验室专注于空气动力学研究，这意味着它从一开始就参与了数值计算。 1955 年，艾姆斯研究中心购置了第一台存储程序电子计算机 IBM 650。1958 年，在中心成为新成立的国家航空航天局 (NASA) 的一部分后不久，它购置了一台 IBM 704，并于 1961 年用 IBM 7090 取代。15^这些计算机全都用于科学计算，例如卫星轨迹、热传递和粒子物理。

The NASA-Ames Research Center, located in Mountain View, California, had been a center for high-speed aerodynamics research since its founding in 1940, as part of what was then the National Advisory Committee for Aeronautics. The shock of the Soviet’s Sputnik in 1957, followed by President Kennedy’s challenge to the nation to put a man on the Moon before the end of the 1960s, gave the center a sense of urgency not seen since the Second World War. Its focus on aerodynamic research meant that the laboratory had been involved with numerical calculations from its beginning. In 1955 Ames had acquired its first stored-program electronic computer, an IBM 650. In 1958, shortly after the center became part of the newly founded National Aeronautics and Space Administration (NASA), it acquired an IBM 704, replaced in 1961 by an IBM 7090.¹⁵ These were all used for scientific calculations, for example, satellite trajectories, heat transfer, and particle physics.

图 4.1

IBM 编码表。程序员不仅几乎看不到计算机，甚至可能看不到将程序输入计算机的键盘。（来源：Thomas E. Bergin。）

Figure 4.1

IBM coding forms. Not only would a programmer hardly ever see the computer, he or she might never even see the keypunch on which the programs were entered into the computer. (Source : Thomas E. Bergin.)

1961 年秋，该中心购买了一台中型 Honeywell H-800 计算机，用于处理风洞数据。随后，该中心又购买了其他类似的机器，用于专门用途：控制实验、操作飞行模拟器和减少风洞数据。这些计算机来自 Honeywell、数字设备公司、科学数据系统 (SDS) 和 EAI。IBM 还推出了 1800 计算机，用于实时控制实验，但 IBM 在中心的主要业务是用于“一般科学”工作的大型集中式计算系统^。16

In the fall of 1961 the center acquired a medium-size Honeywell H-800 for processing of wind tunnel data. That was followed by the acquisition of other similar machines for dedicated purposes: controlling experiments, operating a flight simulator, and reducing wind tunnel data. These computers came from Honeywell, Digital Equipment Corporation, Scientific Data Systems (SDS), and EAI. IBM was also represented by its 1800, which was used for controlling an experiment in real time, but IBM’s presence at the Center was mainly in the large, centralized computing system used for “general scientific” work.¹⁶

整个 20 世纪 60 年代，对中央 IBM 装置的需求以每年超过 100% 的复合率增长。满足这一需求一直是 Ames 计算部门的头疼事。从 1963 年开始，7090 升级为称为直接耦合系统的机器组合。其核心是 7 月收购的 IBM 7094。为了防止这台机器偏离其主要科学工作，IBM 7040（本身就是一台大型主机）与其耦合以处理输入/输出 (I/O)。尽管 7094 有通道，但直接耦合系统允许 7040 处理 I=O，从而减轻 7094 的负担。每台机器的核心内存为 32,768 个 36 位字。IBM 7740 通信计算机处理与多个远程终端的连接。

Throughout the 1960s the demands on the central IBM installation grew at a compounded rate of over 100 percent a year. Meeting that demand was a never-ending headache for the Ames Computation Division. Beginning in 1963 the 7090 was upgraded to a complex of machines called a Direct Couple System. At its heart was an IBM 7094, acquired in July. To keep this machine from being diverted from its main scientific work, an IBM 7040—itself a large mainframe—was coupled to it to handle input/output (I/O). Although the 7094 had channels, the Direct Couple System allowed the 7040 to handle I=O instead, putting less strain on the 7094. Each machine had a core memory of 32,768 36-bit words. An IBM 7740 communications computer handled a connection to several remote terminals.

直接耦合系统主要通过磁带卷与外界通信，磁带卷由连接到键盘打孔机的 IBM 1401 准备；1401 还处理从 7074 传来的磁带的打印结果。1401 还可以作为独立计算机运行，而且是一台非常强大的计算机。NASA 确实以这种方式使用它来处理中心的行政工作，例如预算。17 ¹⁹⁷⁴年，当 DCS 被宣布为政府剩余财产时，完整的 DCS 价值 160 万美元。每月费用约为 35,000 美元。各种其他设备完善了配置。

The Direct Couple System communicated to the outside world mainly through reels of magnetic tape, which were prepared by an IBM 1401 connected to a keypunch; the 1401 also handled printing results, from tapes brought to it from the 7074. The 1401 could also operate as a stand-alone computer, a very capable one at that. NASA did use it this way to handle the center’s administrative work, such as budgeting.¹⁷ In 1974, when it was declared surplus government property, the complete DCS was valued at $1.6 million. Monthly costs were in the range of around $35,000. A variety of other equipment rounded out the configuration.

在太空竞赛的激烈年代，这种直接耦合系统一直服务于艾姆斯实验室。其他大型研究中心也出现了类似的配置，尤其是西海岸的航空航天公司。到 1968 年，该系统全天候工作（周末除外），每月大约有 27 个小时用于维护，并且运行能力远远超出。那一年，IBM System/360 Model 50 取代了它，仅处理器的租金就高达每月 21,000 美元，整个系统的租金为每月 45,000 美元。然而，NASA 仍然不愿意将行政工作从 1401 转移到 360，并最终购买了另一台小型计算机来处理这些任务。尽管如此，360 上安装的主要程序之一是一个软件，它允许它像 1401 一样运行为该机器开发的程序^。18

This Direct Couple System served the Ames laboratory through the dramatic years of the space race. Similar configurations appeared at other large research centers, especially among West Coast aerospace firms. By 1968 the system was working around the clock (except for weekends), with about twenty-seven hours a month reserved for maintenance, and it was running well over capacity. That year an IBM System/360 Model 50 replaced it, at a rental of $21,000 a month for the processor alone, and $45,000 a month for the whole system. NASA was still reluctant to move administrative work off the 1401 to the 360, however, and eventually acquired another small computer to handle those tasks. Nevertheless, one of the main programs installed on the 360 was a piece of software that allowed it to behave as if it were a 1401, to run programs developed for that machine.¹⁸

艾姆斯的大型系统采用批量操作，使用穿孔卡片、磁带和行式打印机。一些小型计算机，尤其是那些连接到风洞、模拟器或其他设备的计算机，实时运行——以与数据呈现相同的速度处理数据。这些系统可能还允许用户直接交互控制。到 20 世纪 60 年代末，艾姆斯的人们希望通过一种称为分时的技术将这种交互式访问扩展到其主机系统。分时将在下一章中详细讨论，但现在值得一提的是它是如何首次出现在 NASA 等地方的^。19

The large systems at Ames operated in batch, using decks of punched cards, tape, and line printers. Some of the smaller computers, especially those connected to a wind tunnel, simulator, or other piece of apparatus, operated in real time—processing data as fast as data were presented to it. These systems might also have a provision for direct, interactive control by a user. By the end of the 1960s people at Ames wanted to extend that kind of interactive access to its mainframe system, through a technique called time-sharing. Time-sharing will be discussed at length in the next chapter, but for now it is worth noting how it first appeared at places like NASA.¹⁹

1969 年，艾姆斯实验室安装了 IBM 360 Model 67，这是 IBM 为该产品线提供分时功能而做出的尝试。但 Model 67 令人失望，²⁰到 1971 年，其“繁重的计算负荷”被转移到其他机器上。Model 67 被保留下来，但重新配置为“通信中心”，主要用于连接新建立的 ARPA 网络。21 Model ⁶⁷的失败是由于其设计缺陷以及难以将分时功能纳入 NASA-Ames 的工作模式。NASA 工程师使用计算机分析风洞数据。在这些问题中，基本程序保持不变，每次测试都会有新数据到达。风洞提前安排好，程序经过调试，工程师们在两个庞大而昂贵的系统（风洞和计算机）之间建立了节奏。由于主计算机的全部资源都用于处理他们每晚生成的数据，因此他们相信第二天早上会得到有用的结果。有了分时系统，情况就不一样了：一个人的工作运行速度取决于其他人在使用机器，以及他们正在做什么样的工作。这是不可预测的，也不在风洞团队的控制范围内。22^在接下来的案例研究中，随着交互式计算越来越成为一种选择，我们将遇到这些技术和社会问题的变体。表 4.1列出了 NASA-Ames 在此期间的计算设施。

In 1969 the Ames laboratory installed an IBM 360 Model 67, IBM’s attempt to provide for time-sharing on that product line. But the Model 67 proved a disappointment,²⁰ and by 1971 its “heavy compute load” was shifted to other machines. The Model 67 was retained but reconfigured to serve as a “communications center,” mainly for connection to the newly established ARPA-Network.²¹ The Model 67’s failure was due to deficiencies in its design and to difficulties in incorporating timesharing into work patterns at NASA-Ames. NASA engineers were using computers to analyze wind-tunnel data. In these problems the basic program remained unchanged, with new data arriving with each test. Wind tunnels were scheduled well in advance, the programs were debugged, and the engineers had established a rhythm between the two large and expensive systems: tunnel and computer. With the full resources of the main computer brought to bear on the data they generated each evening, they were confident that they would have useful results the next morning. With time-sharing that was not the case: how fast or slow one person’s job ran depended on who else was using the machine and what kinds of jobs they were doing. That was unpredic and not under the wind tunnel team’s control.²² In the case studies that follow we shall encounter variants of both these technical and social issues as interactive computing became more of an option. Table 4.1 lists the computing facilities at NASA-Ames during this period.

表 4.1

1955 年至 1969 年 NASA-Ames 的选定计算机安装情况

Table 4.1

Selected computer installations at NASA-Ames, 1955–1969

获取日期	发布日期	设备	使用
2/55	6/63	数据采集器	风洞数据的在线处理。
5/55	9/58	IBM 650	通用科学计算。
9/58	7/61	IBM 704	通用科学计算以及卫星轨迹和热传递计算。取代了 IBM 650。
7/61	3/63	IBM 7090	通用科学计算。增加了入口设计计算和等离子体粒子研究的额外能力。取代了 IBM 704。
11/61		霍尼韦尔 800	风洞数据的在线处理。
3/63	7/64	IBM 7094	取代了 IBM 7090。
7/64		IBM 7094/7040 直接耦合系统 (DCS)	通用科学计算、大规模数据缩减和管理数据处理。包括 IBM 1401。取代了 IBM 7094。
8/64		霍尼韦尔 200	用作 H-800 系统的 I/O 控制单元；取代相关的 HON-800 系统。
7/65	12/67	IBM 7740、4 台 IBM 1440	通信单元和终端提供对中央设施的远程访问。
12/67		IBM 360/50 和 IBM 1800	通用科学计算。采购以补充 DCS 容量并承担远程作业输入功能。取代了 IBM 7740、四台 IBM 1440 和两台 IBM 1401 系统。
1968		伊利亚克IV	先进的并行处理。早期的超级计算机。
7/69		IBM 360/67	分时操作。

来源：数据来自美国国家航空航天局艾姆斯研究中心，“ADPE 收购计划：拟议的中央计算机设施”（1969 年 10 月）：4-5；NASM 档案。

Source : Data from NASA Ames Research Center, “ADPE Acquisition Plan: Proposed Central Computer Facility” (October 1969): 4–5; NASM Archives.

美国国税局

The IRS

美国国税局在处理数据方面承担着如此艰巨的工作，人们几乎无法想象如果没有蓝十字和美国宇航局艾姆斯研究中心的数据，他们的工作将如何进行。与蓝十字和美国宇航局艾姆斯研究中心一样，其对计算机的需求从 1959 年到 1969 年呈数量级增长。我们还将看到，它也遵循了类似的轨迹。

The U.S. Internal Revenue Service performs such a monumental job of processing data that one can scarcely imagine that they ever did their work without As with Blue Cross and NASA-Ames its computer needs increased by orders of magnitude from 1959 to 1969. We shall also see that it followed a similar trajectory.

大多数美国人都熟悉这个机构的基本运作。与 NASA 的工作相比，它主要涉及简单的算术，涉及的金额很少超过一百万。但与科学计算不同，这些计算必须精确到分。这项工作的真正特色是 IRS 必须年复一年地处理大量的纳税申报单，没有空闲时间。

The basic operation of this agency is familiar to most Americans. In contrast to the work of NASA, it involves mainly simple arithmetic and the quantities involved are seldom more than a million. But unlike scientific calculations, these calculations have to be accurate to the penny. What really distinguishes this work is the huge number of tax returns the IRS must process, year in and year out, with no slack time.

为应对第二次世界大战而筹集资金的需要引发了一系列事件，这些事件最终改变了美国国税局的命运。23 当时，需要申报和纳税的美国人数量从大约 800 万增加到 6000 万；从工资中^预扣税款的做法也变得很普遍。国税局使用 Friden 计算器、Burroughs 或 National 会计机以及铅笔和纸张来处理这项工作。直到 1948 年才安装打孔卡设备。1955 年，该机构在堪萨斯城的区域中心安装了一台 IBM 650，在那里它帮助测试性地处理了 110 万份报税表。但打字仍然是每个区域中心的主要活动，大约有 350 名员工输入每份报税表的基本信息^。24

The need to raise revenue to wage the Second World War set in motion events that would transform the Internal Revenue Service.²³ At that time the number of Americans who were required to file returns and pay taxes increased from around eight million to sixty million; the practice of withholding tax from a paycheck also became common. The IRS handled this work with a combination of Friden calculators, Burroughs or National accounting machines, and pencil and paper. Punched card equipment was not installed until 1948. In 1955 the agency installed an IBM 650 in its regional center in Kansas City, where it helped process 1.1 million returns on a test basis. But keypunching was still the main activity at each regional center, where around 350 employees keyed in basic information from each return.²⁴

1959 年，美国财政部授权 IRS 将其业务完全计算机化。在向 40 家制造商征求投标后，IRS 选择了 IBM。每个区域中心都安装了一台 IBM 1401，该计算机具有 4K 核心内存、读卡器、打孔机、行式打印机和两个磁带驱动器。西弗吉尼亚州马丁斯堡的国家中心购买了一台 IBM 7070 大型机（这是多台大型机中的第一台）。（请注意，到目前为止，IRS 走的路与马萨诸塞州蓝十字相同。）电子处理转换在 1967 年完成：Honeywell H-200 取代了 1401，IBM 360 取代了^7070。25

In 1959 the U.S. Treasury Department authorized the IRS to computerize its operations fully. The IRS selected IBM after soliciting bids from forty manufacturers. An IBM 1401 with a 4K core memory, a card reader, punch, line printer, and two tape drives was installed in each regional center. An IBM 7070 mainframe, the first of several, was purchased for a National Center established in Martinsburg, West Virginia. (Note that thus far the IRS was following the same path as Massachusetts Blue Cross.) The changeover to electronic processing was complete by 1967: Honeywell H-200s had replaced the 1401 and IBM 360s had replaced the 7070s.²⁵

尽管现在存储程序计算机正在处理报税表，但流程的第一步仍然是打字员将报税表数据输入打孔卡上 — 到 20 世纪 60 年代中期，每年有超过 4 亿张打孔卡，为超过 1 亿纳税人服务。房间里挤满了女性，她们以稳定、不懈的速度工作，每位女性的眼睛都盯着她左边的报税表，右手悬在打字机上。每个区域中心的 1401 读取卡片，验证所需信息是否存在，进行一些简单的数据缩减，并将结果传输到磁带上。快递员将这些磁带空运到马丁斯堡，在那里 7070 处理报税表。然后，国家中心将磁带发送到财政部以签发退税支票；对于其他不太幸运的人，磁带被送回区域中心以发送账单或以其他方式要求更多信息。由于后面会提到计算机网络这个话题，因此值得注意的是，携带这些磁带的快递员在一次跨国飞行中以每秒约 30,000 比特的速度传输数据。这相当于 20 世纪 90 年代的个人电脑通过普通电话线可以处理的数据量^。26

Although stored-program computers were now processing the returns, the first step in the process was still keypunchers entering data from returns onto punched cards—over 400 million cards a year, for over 100 million taxpayers, by the mid-1960s. Rooms full of mostly women worked at a steady, unflagging pace, each woman’s eye focused on a return propped up to her left, her right hand floating over a keypunch. The 1401s at each regional center read the cards, verified that the required information was there, did some simple data reduction, and transferred the results to tape. Couriers flew these tapes to Martinsburg, where the 7070 processed the returns. The National Center then sent a tape to the Treasury Department to issue refund checks; for others less fortunate, a tape was sent back to a regional center to send out a bill or otherwise ask for more information. Since the topic of computer networking will arise later, it is worth noting that a courier carrying these reels of tape was moving data at a rate of about 30,000 bits per second on a cross-country flight. That was equivalent to what a personal computer in the 1990s could handle over ordinary telephone lines.²⁶

到 1965 年，国税局开始用一个独特的数字来识别每个纳税人——他或她的社会安全号码——从而消除了处理同名人员的混乱。这需要国会的一项法案；人们很容易忘记社会安全号码在 20 世纪 30 年代建立时的不起眼的起源。要求人们在所有表格上填写这个号码，再加上随之而来的关于“电子大脑”的宣传，导致国会议员收到了几封关于“老大哥”的恶毒信件。很少有人意识到他们刚刚跨越了社会分水岭。（他们后来会意识到，正如我们将讨论的那样。）打卡在 1967 年结束，当时安装了允许将数据直接输入到鼓（后来是磁盘）上的机器，但除此之外，现场中心和国家中心之间的这种分工一直持续到 20 世纪 90 年代。（当键盘打孔机退役后，管理人员发现生产率并没有像他们预期的那样提高。通过重新引入一些丢失的声音，操作员能够重新建立保持高数据输入率所需的节奏。）²⁷

By 1965 the IRS was identifying each taxpayer by a unique number— his or her Social Security number—eliminating the confusion of handling persons with the same names. That had required an act of Congress; one easily forgets the modest origins of the Social Security number when it was established in the 1930s. Requiring one to put this number on all forms, plus the attendant publicity about “electronic brains,” led to a few nasty letters to congresspeople about “Big Brother.” Few realized the social watershed they had just crossed. (They would later on, as will be discussed.) The punching of cards ended in 1967, when machines were installed that allowed direct entry of data onto a drum (later a disk), but otherwise this division of labor among field centers and the National Center remained into the 1990s. (When the keypunch machines were retired, managers found that productivity did not go up as they expected. By reintroducing some of the sound that was lost, the operators were able to reestablish the rhythm necessary to maintain high rates of data entry.)²⁷

与 Blue Cross 和 NASA-Ames 一样，IRS 按顺序处理数据。要查找或更改特定记录，需要安装相应的磁带，并通过磁带驱动器运行，直到出现该记录。有问题的纳税人必须等待其记录被送到区域中心。到 20 世纪 70 年代中期，该操作已形成一种模式，即主文件每周更新一次，生成一个微缩胶片上的更新文件，该文件可以空运到区域中心以解决有关特定申报表的问题。这种数据检索不是由于 IRS 的任何官僚惰性；它是系统结构的一部分。

As at Blue Cross and NASA-Ames, the IRS processed data sequentially. To find or change a particular record, one mounted the appropriate tape and ran it through a tape drive until that record appeared. A taxpayer who had a problem would have to wait for his or her record to be delivered to a regional center. By the mid-1970s the operation had settled into a pattern whereby the master file was updated once a week, producing an updated file on microfilm that could be flown to a regional center to address questions about a specific return. This kind of data retrieval was not due to any bureaucratic inertia on the part of the IRS; it was built into the structure of the system.

到 1967 年，随着计算机化处理到位并顺利运行，该机构开始寻找改进的系统。它不仅希望消除打卡，还希望完全消除手动输入数据——无论是使用可以读取手写数字的机器，还是让纳税人以某种机器可读的形式填写报税表。该机构还打算从顺序访问数据转向随机、在线访问数据。

By 1967, with the computerized processing in place and operating smoothly, the agency began looking for an improved system. It hoped not only to eliminate punching cards but also to eliminate manually keying in data altogether—whether by using machines that could read handwritten figures, or by having the taxpayer fill out his or her return in some sort of machine-readable form. The agency also intended to move away from sequential toward random, on-line access to data.

为实现这些目标，政府制定了一项雄心勃勃的计划，称为税务管理系统 (TAS)，预计耗资 6.5 亿至 8.5 亿美元。28^税务处理将分散到全美十个服务中心，而不仅仅是马丁斯堡的一个服务中心。政府将安装一个“采用分散数据库的批量和实时交易导向型计算机网络”，通过 8,000 多个终端之一即可直接访问纳税人信息。29^纳税人数据将存储在中心的“随机存取存储设备”（可能是磁盘）上，而不是磁带上。其他终端将允许数据直接输入网络，而无需打卡。

An ambitious plan, projected to cost 650 to 850 million dollars, called the Tax Administration System (TAS), was conceived to implement these goals.²⁸ Processing would be dispersed to ten service centers across the United States, instead of only to the one in Martinsburg. A combination “batch and realtime [sic] transaction-oriented computer network employing a decentralized database” would be installed, with direct access to taxpayer information available at one of over 8,000 terminals.²⁹ Taxpayer data would be stored at the centers on “random access storage devices” (probably magnetic disks), instead of on tapes. Other terminals would allow data entry directly into the network, without the need for punching cards.

TAS 的规划者们花了很多心思来确保系统的安全——免受物理损坏、恶意入侵和简单的人为错误。但是，对计算机的不信任的种子已经萌芽。20 世纪 60 年代末，许多公民对联邦政府的诚实性产生了质疑。在导致尼克松总统于 1974 年辞职的水门事件听证会上，人们发现白宫已经破坏了国税局精心建立的诚信之墙，该墙旨在保护其业务不受政治干预。尽管国税局尚未采用交互式数据检索系统，但白宫特工已经能够获得那些不利于他们的人的税务记录。信任已经消失，没有信任，任何税收制度都无法运转。

The planners of the TAS gave much thought to making the system secure—from physical damage, from malicious intrusion, and from simple human errors. But the seed of mistrust in computers that had lain dormant now sprouted. The late 1960s was a time when many citizens questioned the federal government’s truthfulness. During the Watergate hearings, which led to President Nixon’s resignation in 1974, it was revealed that the White House had breached the wall of integrity that the IRS had carefully built up to shield its operations from political interference. Although the IRS had not yet adopted an interactive data-retrieval system, White House operatives had been able to obtain the tax records of those not in their favor. Trust, without which no system of taxation can function, had eroded.

这次国会收到了不少愤怒的信件。国会指示总审计署 (GAO) 研究拟议的 TAS 对隐私的影响；GAO 于 1976 年和 1977 年初发布的初步报告批评该系统没有充分解决安全和隐私问题。30^一份报告被泄露给行业杂志《计算机世界》，该杂志刊登了一篇头条新闻，标题为“拟议的 IRS 系统可能对隐私构成威胁”。31 ¹⁹⁷⁷年春，参议院和众议院举行了听证会，会上 IRS 官员被问及一些问题，例如是否有人可以“将终端与现有终端网络并行连接”，从而能够“提取所有这些信息”。32^一些IRS 员工回忆说，国会议员决定该机构可以竞标哪些类型的计算机架构^。33

This time there were more than a few irate letters to Congress. Congress directed the General Accounting Office (GAO) to look at the privacy implications of the proposed TAS; the GAO’s preliminary report, issued in 1976 and early 1977, criticized the system for not addressing security and privacy issues thoroughly enough.³⁰ A copy of the report was leaked to the trade journal Computerworld, which ran a lead story under the headline “Proposed IRS System May Pose Threat to Privacy.”³¹ In the spring of 1977 there were hearings in both the Senate and the House, at which IRS officials were asked questions such as whether someone could “attach a terminal in parallel” with the existing terminal network and thereby be capable of “pulling all this information out.”³² Some IRS employees recall members of Congress dictating what types of computer architecture the agency was allowed to bid on.³³

在国会的压力下，国税局于 1978 年 1 月放弃了 TAS 计划。取而代之的是，国税局提出了一项“设备更换和增强计划”。（国会迫使他们删除了“增强”一词。）保留了旧的架构，即在磁带上保存集中式主文件。国税局的帕特里克·拉特尔称这是“以非常安全的方式迈向未来的方式”。34^{即时在线访问记录是被禁止的。由于国会的敌意，该机构}^只能勉强维持下去。1985 年，该系统崩溃了；报纸刊登了退税被扔进垃圾箱、退税支票丢失等耸人听闻的故事。35国会改变了主意，批准拨款开发新的数据处理架构。

Under pressure from Congress, the IRS dropped its plans for the TAS in January 1978. In its place the IRS proposed an “equipment replacement and enhancement program.” (Congress made them drop the word “enhancement.”) The old architecture in which a centralized master file was kept on magnetic tape was retained. Patrick Ruttle of the IRS called this “a way of moving into the future in a very safe fashion.”³⁴ Instantaneous on-line access to records was verboten. Hamstrung by a hostile Congress, the agency limped along. In 1985 the system collapsed; newspapers published lurid stories of returns being left in dumpsters, refund checks lost, and so on.³⁵ Congress had a change of heart and authorized money to develop a new data-handling architecture.

美国宇航局载人航天计划

NASA’s Manned Space Program

美国宇航局艾姆斯研究中心和美国国税局都曾尝试摆脱批处理和顺序访问数据的方式，但都失败了，至少在最初是这样。但这些失败揭示了批处理操作的优势，而这些优势可能被忽略了。批处理操作保留了早期制表机时代的社会背景的连续性；多年来，它还经过了微调，让客户能够以最少的钱充分利用机器。批处理的真正问题更多的是哲学问题，而不是技术或经济问题。它使计算机相当于无马马车或无线电报——它比制表机或手工计算运行速度更快，处理的数量也更大，但它并没有改变工作的性质。

Both NASA-Ames and the IRS made attempts to move away from batch processing and sequential access to data, and both failed, at least at first. But the failures revealed advantages of batch operation that may have been overlooked otherwise. Batch operation preserved continuity with the social setting of the earlier tabulator age; it also had been fine-tuned over the years to give the customer the best utilization of the machine for his or her dollar. The real problem with batch processing was more philosophical than technical or economic. It made the computer the equivalent of a horseless carriage or wireless telegraph—it worked faster and handled greater quantities than tabulators or hand calculations, but it did not alter the nature of the work.

在此期间，直到 20 世纪 60 年代末，只有在成本不是问题的情况下，才能实现直接、交互式地访问计算机。NASA 的载人航天计划就是这样一个项目，它使用与 IRS、NASA-Ames 和 Blue Cross 相同的硬件开发了这种访问方式。36 ²⁰世纪 50 年代末，一个不考虑成本的项目开始了：美国竞相在 2020 年前将人类送上月球。

During this period, up to the late 1960s, direct, interactive access to a computer could exist only where cost was not a factor. NASA’s Manned Space Program was such an installation where this kind of access was developed, using the same kind of hardware as the IRS, NASA-Ames, and Blue Cross.³⁶ In the late 1950s a project was begun for which cost was not an objection: America’s race to put men on the Moon by the end of the decade.

大部分太空任务都是在无动力飞行中滑行。在发射的最初几分钟内，当发动机燃烧时，必须进行大量计算。如果飞船偏离了航线，必须将其摧毁，以防止其撞上人口稠密的地区。如果发射顺利，必须快速计算最终轨道以确定其是否稳定，并且必须将这些信息传输到全球各地的跟踪站。计算工作量巨大，必须在几秒钟内完成。

Most of a space mission consists of coasting in unpowered flight. A lot of computing must be done during the initial minutes of a launch, when the engines are burning. If the craft is off-course, it must be destroyed to prevent its hitting a populated area. If a launch goes well, the resulting orbit must be calculated quickly to determine if it is stable, and that information must be transmitted to tracking stations located around the globe. The calculations are formidable and must be carried out, literally, in a matter of seconds.

1957 年，海军研究实验室在华盛顿特区为先锋计划建立了一个控制中心，这是美国首次尝试将卫星送入轨道。该中心希望实时将有关卫星的信息传送到 IBM 704 计算机中：以尽可能快的速度计算出轨道，只要助推器和卫星的遥测数据能够输入到计算机中。37^他们没有实现这个目标——数据仍然必须打到卡片上。1960 年 11 月，NASA 在马里兰州格林贝尔特新成立的戈达德太空飞行中心安装了一套由两台 7090 计算机组成的系统。对于此安装，实现了实时处理。每台 7090 都可以实时计算轨道，其中一台作为另一台的备份。发射数据在卡纳维拉尔角收集并传输到格林贝尔特；使用一台 IBM 709 的备份系统位于百慕大，这是火箭发射后将经过的第一块陆地。世界各地还建立了其他雷达站，以提供持续覆盖^。38

In 1957 the Naval Research Laboratory established a control center in Washington, D.C., for Project Vanguard, America’s first attempt to orbit a satellite. The Center hoped to get information about the satellite to its IBM 704 computer in real time: to compute a trajectory as fast as the telemetry data about the booster and satellite could be fed to it.³⁷ They did not achieve that goal—data still had to be punched onto cards. In November 1960 NASA installed a system of two 7090 computers at the newly formed Goddard Space Flight Center in Greenbelt, Maryland. For this installation, real-time processing was achieved. Each 7090 could compute trajectories in real time, with one serving as a backup to the other. Launch data were gathered at Cape Canaveral and transmitted to Greenbelt; a backup system, using a single IBM 709, was located in Bermuda, the first piece of land the rocket would pass over after launch. Other radar stations were established around the world to provide continuous coverage.³⁸

系统计算出预测轨迹并将其传回佛罗里达州的 NASA 任务控制中心。根据该轨迹是否与计划一致，飞行控制员做出“继续”或“不继续”的决定，从发动机关闭后十秒开始，并在整个任务期间间歇进行。39^发射时，专用的 Atlas Guidance 计算机以每秒 1,000 比特的速度处理数据。发动机关闭后，数据以每秒 6 个字符的速度流入戈达德计算机。40^对于记得 1962 年 2 月约翰·格伦的轨道飞行的那一代美国人来说，水星控制官定期发出简洁的“进入轨道！”声明是这次飞行中最激动人心的部分之一。

The system calculated a predicted trajectory and transmitted that back to NASA’s Mission Control in Florida. Depending on whether that trajectory agreed with what was planned, the flight controller made a “Go” or “No Go” decision, beginning ten seconds after engine cut-off and continuing at intervals throughout the mission.³⁹ At launch, a special-purpose Atlas Guidance computer handled data at rates of 1,000 bits per second. After engine cut-off the data flowed into the Goddard computers at a rate of six characters a second.⁴⁰ For the generation of Americans who remember John Glenn’s orbital flight in February 1962, the clipped voice of the Mercury Control Officer issuing periodic, terse “Go for orbit!” statements was one of the most dramatic aspects of the flight.

在典型的 7090 安装中，其通道处理中央处理器与计算机房中的外围设备之间的输入和输出。在这种情况下，数据来自距离格林贝尔特一千英里的佛罗里达州的雷达站。IBM 和 NASA 开发了通道增强功能，进一步调节和处理数据。他们还开发了名为 Mercury Monitor 的系统软件，允许某些输入数据中断处理器正在执行的任何操作，以确保不会忽略危及生命的情况。就像一位忙碌的高管，其备忘录被标记为紧急、非常紧急和极其紧急，允许多个优先级别，由特殊的“陷阱处理器”指示。在执行“陷阱”时，系统首先保存计算机寄存器的内容，以便在处理中断后返回这些数据^。41

In a typical 7090 installation, its channels handled input and output between the central processor and the peripheral equipment located in the computer room. In this case the data was coming from radar stations in Florida, a thousand miles away from Greenbelt. IBM and NASA developed an enhancement to the channels that further conditioned and processed the data. They also developed system software, called Mercury Monitor, that allowed certain input data to interrupt whatever the processor was doing, to ensure that a life-threatening situation was not ignored. Like a busy executive whose memos are labeled urgent, very urgent, and extremely urgent, multiple levels of priority were permitted, as directed by a special “trap processor.” When executing a “trap,” the system first of all saved the contents of the computer’s registers, so that these data could be returned after the interruption was handled.⁴¹

Mercury Monitor 代表着远离批量操作的重要一步，展示了使用未设计为批量操作的商用大型机可以完成哪些工作。42^它发展成为 IBM 最雄心勃勃、最成功的软件产品之一，为该公司进入后来用于银行、航空预订系统和大型在线数据网络的在线系统奠定了基础^。43

The Mercury Monitor represented a significant step away from batch operation, showing what could be done with commercial mainframes not designed to operate that way.⁴² It evolved into one of IBM’s most ambitious and successful software products and laid the foundation for the company’s entry into on-line systems later adopted for banking, airline reservations systems, and large on-line data networks.⁴³

20 世纪 60 年代中期，任务控制中心迁至休斯顿，安装了一套由三台（后来为五台）7094 计算机组成的系统，每台计算机都连接到一台 IBM 1401。1966 年 8 月，7094 被基于 IBM 360 Model 75 的系统所取代。简单的 Mercury Monitor 已发展成为标准 IBM 360 操作系统的实时扩展。IBM 工程师 Tom Simpson、Bob Crabtree 和其他三人将该程序称为 HASP（休斯顿自动假脱机优先级 - SPOOL 本身是早期的缩写）。它允许 Model 75 同时作为批处理和实时处理器运行。该系统被证明是有效的，对于一些客户来说，它比 IBM 的标准 System/360 操作系统更受欢迎。HASP 很快被其他商业设施采用，并在 20 世纪 70 年代成为完全支持的 IBM 产品^。44

In the mid-1960s Mission Control moved to Houston, where a system of three (later five) 7094 computers, each connected to an IBM 1401, was installed. In August 1966 the 7094s were replaced by a system based on the IBM 360, Model 75. The simple Mercury Monitor had evolved into a real-time extension of the standard IBM 360 operating system. IBM engineers Tom Simpson, Bob Crabtree and three others called the program HASP (Houston Automatic Spooling Priority—SPOOL was itself an acronym from an earlier day). It allowed the Model 75 to operate both as a batch and real-time processor. This system proved effective and for some customers was preferred over IBM’s standard System/360 operating system. HASP was soon adopted at other commercial installations and in the 1970s became a fully supported IBM product.⁴⁴

如果没有阿波罗任务的独特性，IBM 大型机的这些修改就不可能发生：其目标（将人类送上月球并安全返回）和最后期限（“十年内”）。对于大多数其他客户来说，这种修改既不切实际，也不受 IBM 允许，这些客户通常租用设备，并不拥有设备。45 ^NASA的修改确实表明，大型商用大型机可以以批处理模式以外的方式运行。NASA 的解决方案涉及大量硬件和软件的定制工作，但随着时间的推移，其他更传统的客户能够基于这项工作构建类似的系统。

These modifications of IBM mainframes could not have happened without the unique nature of the Apollo mission: its goal (to put a man on the Moon and return him safely) and its deadline (“before the decade is out”). Such modifications were neither practical nor even permitted by IBM for most other customers, who typically leased and did not own equipment.⁴⁵ NASA’s modifications did show that a large, commercial mainframe could operate in other than a batch mode. NASA’s solution involved a lot of custom work in hardware and software, but in time other, more traditional customers were able to build similar systems based on that work.

小型计算机

The Minicomputer

我们已经从上至下描述了计算领域的变化，这些变化是由资金雄厚的客户不断增长的需求引起的，现在我们将看看这些变化是如何受到固态物理、电子学和计算机体系结构研究进展的影响的。结果是一种称为“小型计算机”的新型机器。它并不是大型机或使用大型机文化的直接竞争对手。相反，小型计算机开辟了全新的应用领域。它的发展是一种文化、经济和技术现象。它使大量人群（最初是工程师和科学家，后来是其他人）能够直接与计算机交互。小型计算机，特别是由电传打字机操作的小型计算机，引入了计算机作为个人交互设备的概念。最终，随着小型计算机让位于其后代个人计算机，这一概念将改变我们的文化并主导我们的期望。

Having described changes in computing from the top down, changes caused by increased demands by well-funded customers, we’ll now look at how these changes were influenced by advances in research into solid-state physics, electronics, and computer architecture. The result was a new type of machine called the “minicomputer.” It was not a direct competitor to mainframes or to the culture of using mainframes. Instead the minicomputer opened up entirely new areas of application. Its growth was a cultural, economic, and technological phenomenon. It introduced large groups of people—at first engineers and scientists, later others—to direct interaction with computing machines. Minicomputers, in particular those operated by a Teletype, introduced the notion of the computer as a personal interactive device. Ultimately that notion would change our culture and dominate our expectations, as the minicomputer yielded to its offspring, the personal computer.

建筑学

Architecture

小型计算机由多种因素决定：架构、封装、第三方在开发应用程序方面的作用、价格和融资。值得详细讨论第一个因素，即架构，以了解小型计算机与当时流行的计算机有何不同。

A number of factors define the minicomputer: architecture, packaging, the role of third-parties in developing applications, price, and financing. It is worth discussing the first of those, architecture, in some detail to see how the minicomputer differed from what was prevalent at the time.

20 世纪 60 年代早期，典型的 IBM 大型机每次以 36 位为单位进行操作，使用中央处理器中的一个或多个寄存器。其他寄存器处理寻址、索引以及两个 36 位数相乘时生成的额外数字。计算机最快、最复杂、最昂贵的电路都在这里。较短的字长可以降低复杂性，从而降低成本，但这会带来一些不利影响。最大的不利影响是较短的字长无法在指令中提供足够的位来指定足够的内存地址。这就像试图用七位电话号码但没有区号来提供全国电话服务一样。使用短字的另一个不利影响是，算术运算无法为除最简单的算术之外的任何运算提供足够的数字，除非对机器进行编程以“双精度”运行。IBM 7090 系列中使用的 36 位字相当于十进制数字。这对于大多数应用来说已经足够了，但许多人认为客户不会想要一台无法处理这么多数据的机器。

A typical IBM mainframe in the early 1960s operated on 36 bits at a time, using one or more registers in its central processor. Other registers handled the addressing, indexing, and the extra digits generated during a multiplication of two 36-bit numbers. The fastest, most complex, and most expensive circuits of the computer were found here. A shorter word length could lower the complexity and therefore the cost, but that incurred several penalties. The biggest penalty was that a short word length did not provide enough bits in an instruction to specify enough memory addresses. It would be like trying to provide telephone service across the country with seven-digit phone numbers but no area codes. Another penalty of using a short word was that an arithmetic operation could not provide enough digits for anything but the simplest arithmetic, unless one programmed the machine to operate in “double precision.” The 36-bit word used in the IBM 7090 series gave the equivalent of ten decimal digits. That was adequate for most applications, but many assumed that customers would not want a machine that could not handle at least that many.

小型计算机找到了解决这些缺点的方法。他们通过使计算机的指令代码更复杂来做到这一点。除了指令中指定的操作代码和内存地址外，小型计算机还使用代码的几位来指定扩展内存空间的不同“模式”。一种操作模式可能不直接引用内存位置，而是引用存储所需内存位置的另一个寄存器。这当然增加了复杂性；以双精度操作也很复杂，两者都可能降低计算机的速度。但随着 20 世纪 50 年代末新晶体管的上市，人们可以设计一种处理器，即使增加了这些复杂性，也能保持简单、廉价和快速。

Minicomputers found ways to get around those drawbacks. They did that by making the computer’s instruction codes more complex. Besides the operation code and memory address specified in an instruction, minicomputers used several bits of the code to specify different “modes” that extend the memory space. One mode of operation might not refer directly to a memory location but to another register in which the desired memory location is stored. That of course adds complexity; operating in double precision also is complicated, and both might slow the computer down. But with the newly available transistors coming on the market in the late 1950s, one could design a processor that, even with these added complexities, remained simple, inexpensive, and fast.

Whirlwind 的字长只有 16 位，但商用小型计算机的故事实际上始于一位与超大型计算机相关的发明家：西摩·克雷。1957 年，控制数据公司在双城创立，创始人是威廉·诺里斯，他也是工程研究协会（后来成为雷明顿·兰德 UNIVAC 的一部分，如第 1 章所述）。诺里斯说服了众多与他一起共事的工程师，其中就有克雷。在 UNIVAC 工作期间，克雷曾参与海军战术数据系统 (NTDS) 的研发，这是一台为海军舰艇设计的计算机，也是第一批量产的晶体管机器之一。46 ¹⁹⁶⁰年左右，CDC 推出了其型号 1604，这是一款面向科学客户的大型计算机。此后不久，该公司推出了由 Cray 设计的 160（据 CDC 的一名员工称，“几乎是事后才想到的”），用于处理 1604 的输入和输出。对于 160，Seymour Cray 保留了他为海军系统开创的一些关键功能，尤其是其紧凑的包装。事实上，这台计算机足够小，可以放在一张普通的金属桌子上——偶然发现它的人甚至不知道这是一台计算机。

The Whirlwind had a word length of only 16 bits, but the story of commercial minicomputers really begins with an inventor associated with very large computers: Seymour Cray. In 1957, the Control Data Corporation was founded in the Twin Cities by William Norris, the cofounder of Engineering Research Associates, later part of Remington Rand UNIVAC, as mentioned in chapter 1. Among the many engineers Norris persuaded to go with him was Cray. While at UNIVAC Cray had worked on the Navy Tactical Data System (NTDS), a computer designed for Navy ships and one of the first transistorized machines produced in quantity.⁴⁶ Around 1960 CDC introduced its model 1604, a large computer intended for scientific customers. Shortly thereafter the company introduced the 160, designed by Cray (“almost as an afterthought,” according to a CDC employee) to handle input and output for the 1604. For the 160 Seymour Cray carried over some key features he pioneered for the Navy system, especially its compact packaging. In fact, the computer was small enough to fit around an ordinary-looking metal desk—someone who chanced upon it would not even know it was a computer.

160 通过使用短字长（12 位）以及超出短地址字段限制的内存访问方式，开辟了新局面。47^它能够直接寻址八千字的主内存，并且时钟周期相当快（内存访问时间为 6.4 微秒）。而且 160 的生产成本低廉。当 CDC 以 60,000 美元的价格提供独立版本 160A 时，它找到了现成的市场。控制数据公司正致力于开发高性能机器（后来被称为“超级计算机”，Cray 也因此而闻名），但它并不介意同时销售 160A。Seymour Cray 发明的实际上是一台小型计算机^。48

The 160 broke new ground by using a short word length (12 bits) combined with ways of accessing memory beyond the limits of a short address field.⁴⁷ It was able to directly address a primary memory of eight thousand words, and it had a reasonably fast clock cycle (6.4 microseconds for a memory access). And the 160 was inexpensive to produce. When CDC offered a stand-alone version, the 160A, for sale at a price of $60,000, it found a ready market. Control Data Corporation was concentrating its efforts on very high performance machines (later called “supercomputers,” for which Cray became famous), but it did not mind selling the 160A along the way. What Seymour Cray had invented was, in fact, a minicomputer.⁴⁸

几乎立刻，新的市场开始向不受大型机文化束缚的计算机开放。第一批客户之一，即 Scantlin Electronics, Inc. (SEI) 的负责人 Jack Scantlin，很好地说明了此类设计的潜力。当他在 1962 年看到 CDC 160A 时，他构想了一个围绕它构建的系统，该系统将纽约证券交易所的在线报价提供给全国各地的经纪人。到 1963 年，SEI 的 Quotron II 系统投入运营，在大约 15 秒内提供股票价格，当时纽约证券交易所每天平均交易约 380 万股。49^名SEI 工程师采用了一些巧妙的技巧，将有关股票价格的所有必要信息存储在少量的 12 位字中，但最终证明这台机器（实际上是连接到一个公共内存的两个 160A）完全能够支持这种复杂的应用程序。

Almost immediately new markets began to open for a computer that was not tied to the culture of the mainframe. One of the first customers, which provides a good illustration of the potential of such designs, was Jack Scantlin, the head of Scantlin Electronics, Inc. (SEI). When he saw a CDC 160A in 1962, he conceived of a system built around it that would provide on-line quotations from the New York Stock Exchange to brokers across the country. By 1963 SEI’s Quotron II system was operational, providing stock prices within about fifteen seconds, at a time when trading on the NYSE averaged about 3.8 million shares a day.⁴⁹ SEI engineers resorted to some ingenious tricks to carry all the necessary information about stock prices in a small number of 12-bit words, but ultimately the machine (actually, two 160As connected to a common memory) proved fully capable of supporting this sophisticated application.

数字设备公司

The Digital Equipment Corporation

1957 年，在 CDC 成立的同一年，肯尼斯·奥尔森 (Kenneth H. Olsen) 和哈兰·安德森 (Harlan Anderson) 成立了数字设备公司 (DEC，发音为“deck”)。资金来自美国研究与开发公司，该公司由哈佛商学院教授乔治·多里奥特 (Georges Doriot) 创立，他的目标是找到一种方法，将他在二战期间担任美国陆军军官时观察到的科学和技术创新商业化。他们在马萨诸塞州梅纳德阿萨贝特河对岸的一家羊毛厂的一角开展业务。在麻省理工学院 (MIT) 读书时，奥尔森曾参与为 Whirlwind 安装核心内存以取代其易碎且不可靠的存储管，并在 20 世纪 50 年代中期在列克星敦郊区的麻省理工学院林肯实验室工作。当 IBM 为 SAGE 防空系统制造计算机时，他曾代表林肯实验室与 IBM 合作。1955 年，奥尔森负责林肯实验室的一台名为 TX-0 的计算机，这是一台非常早期的晶体管机器。⁵⁰在他的监督下，TX-0 于 1956 年首次在林肯实验室运行^。51

In the same year that CDC was founded, 1957, Kenneth H. Olsen and Harlan Anderson founded the Digital Equipment Corporation (DEC, pronounced “deck”). Financing came from the American Research and Development Corporation, a firm set up by Harvard Business School Professor Georges Doriot, whose goal was to find a way to commercialize the scientific and technical innovations he had observed during the Second World War as an officer in the U.S. Army. They set up operations in a corner of a woolen mill astride the Assabet River in Maynard, Massachusetts. As a student at MIT, Olsen had worked on fitting the Whirlwind with core memory in place of its fragile and unreliable storage tubes, and in the mid-1950s he had worked for MIT’s Lincoln Laboratory in suburban Lexington. He had represented the Lincoln Lab to IBM when it was building computers for the SAGE air-defense system. In 1955 Olsen had taken charge of a computer for Lincoln Lab called TX-0, a very early transistorized machine.⁵⁰ Under his supervision, the TX-0 first operated at Lincoln Lab in 1956.⁵¹

TX-0 的字长较短，只有 18 位。它被设计用来利用当时由 Philco 生产的新型表面势垒晶体管（它使用了大约 3,600 个）。这些晶体管的速度和质量都比以前任何晶体管都要快得多。尽管每个晶体管的价格为 40 到 80 美元（而电子管的价格约为 3 到 10 美元），而且它们的长期可靠性尚不清楚，但 TX-0 的设计人员很快就发现这些晶体管是可靠的，不需要与其他组件不同的处理。52 ^TX -0 反映了它与交互式 SAGE 系统的联系，它有一个阴极射线管显示器和一支光笔，允许操作员在程序运行时直接与程序交互。该显示器的设计者是 Ben Gurley，他于 1959 年离开林肯实验室，成为数字设备公司的首批员工之一。

The TX-0 had a short word length of 18 bits. It was designed to utilize the new surface-barrier transistors just then being produced by Philco (it used around 3,600 of them). These transistors were significantly faster and of higher quality than any transistors available previously. Although each one cost $40 to $80 (compared to about $3 to $10 for a tube), and their long-term reliability was unknown, the TX-0 designers soon learned that the transistors were reliable and did not need any treatment different from other components.⁵² Reflecting its connections to the interactive SAGE system, the TX-0 had a cathode-ray tube display and a light-pen, which allowed an operator to interact directly with a program as it was running. The designer of that display was Ben Gurley, who left Lincoln Labs in 1959 to become one of Digital Equipment Corporation’s first employees.

TX-0 于 1957 年完工，是世界上最先进的计算机之一。1959 年，数字设备公司推出了由 Gurley 设计的 PDP-1，它采用了 TX-0 的许多架构和电路创新。回想一下，IBM 7090 是一台晶体管机器，采用与真空管 709 相同的架构，用晶体管取代单个电子管。PDP-1 的设计与电子管设计毫无关联；它从一开始就打算充分利用晶体管的优势。它每秒可以进行 100,000 次加法运算，速度不如 IBM 7090，但比同价位的基于鼓的计算机快得多，而且相当不错。它的基本核心内存可容纳 4,000 个 18 位字，后来扩展到 64,000 个。

When completed in 1957, the TX-0 was one of the most advanced computers in the world, and in 1959 when Digital Equipment Corporation offered its PDP-1 designed by Gurley, it incorporated many of the TX-0’s architectural and circuit innovations. Recall that the IBM 7090 was a transistorized machine that employed the same architecture as the vacuum tube 709, with transistors replacing the individual tubes. The PDP-1 owed nothing to tube design; it was intended to take full advantage of what transistors had had to offer from the start. It was capable of 100,000 additions per second, not as fast as the IBM 7090, but respectable and much faster than the drum-based computers in its price class. Its basic core memory held four thousand, later expanded to sixty-four thousand, 18-bit words.

PDP-1 并非 TX-0 的完全复制品，但它模仿了 TX-0 最具创新性的架构特征之一：放弃使用大型机使用的通道，允许 I/O 直接从 I/O 设备传输到核心内存本身。通过精心设计和熟练的编程，这实现了快速 I/O，对中央处理器的运行影响很小，成本和复杂性仅为使用通道的机器的一小部分。53^这种“直接内存访问”(DMA) 以某种形式融入了几乎所有后续 DEC 产品中，并定义了小型计算机的架构。它也内置于现代个人计算机中使用的微处理器中。为了允许这种访问，处理器允许中断在多个级别（最多 16 个）发生，并有专门的电路以正确的顺序处理它们。成本节省是惊人的：正如 DEC 工程师后来所描述的那样，“单个 IBM 通道比 PDP-1 更贵。” ⁵⁴最初的售价为 120,000 美元。

The PDP-1 was not an exact copy of the TX-0, but it did imitate one of its most innovative architectural features: foregoing the use of channels, which mainframes used, and allowing I/O to proceed directly from an I/O device to the core memory itself. By careful design and skillful programming, this allowed fast I/O with only a minimal impact on the operation of the central processor, at a fraction of the cost and complexity of a machine using channels.⁵³ In one form or another this “direct memory access” (DMA) was incorporated into nearly all subsequent DEC products and defined the architecture of the minicomputer. It is built into the microprocessors used in modern personal computers as well. To allow such access to take place, the processor allowed interrupts to occur at multiple levels (up to sixteen), with circuits dedicated to handling them in the right order. The cost savings were dramatic: as DEC engineers later described it, “A single IBM channel was more expensive than a PDP-1.”⁵⁴ The initial selling price was $120,000.

数字设备公司销售了大约 50 台 PDP-1。这算不上是商业上的成功，但它凭借其架构创新在计算机史上占有一席之地，这些创新与约翰·冯·诺依曼 1945 年关于 EDVAC 的报告中体现的创新一样深刻和持久。

Digital Equipment Corporation sold about fifty PDP-1s. It was hardly a commercial success, but it deserves a place in the history of computing for its architectural innovations—innovations that were as profound and long-lasting as those embodied in John von Neumann’s 1945 report on the EDVAC.

PDP-1 的适度销售为 DEC 的下一步奠定了基础。即建立与 IBM 及其竞争对手截然不同的供应商和客户之间的密切关系。从成立之初，IBM 的政策就是租赁而不是出售其设备。该政策使其比竞争对手具有许多优势；它还需要 DEC 所没有的资本资源。尽管 IBM 同意根据 1956 年 1 月生效的同意令出售其机器，但租赁仍然是其首选的经营方式。55^该政策意味着客户场所内的机器不是他或她可以随意处置的；它属于 IBM，只有 IBM 才被允许对其进行修改。上文描述的 NASA 在其休斯顿中心所做的修改是该政策的罕见例外。

The modest sales of the PDP-1 set the stage for Digital’s next step. That was to establish a close relationship between supplier and customer that differed radically from those of IBM and its competitors. From the time of its founding, IBM’s policy had been to lease, not sell, its equipment. That policy gave it a number of advantages over its competitors; it also required capital resources that DEC did not have. Although IBM agreed to sell its machines as part of a Consent Decree effective January 1956, leasing continued to be its preferred way of doing business.⁵⁵ That policy implied that the machine on the customer’s premises was not his or hers to do with as he wished; it belonged to IBM, and only IBM was allowed to modify it. The kinds of modifications that NASA made at its Houston center, described above, were the rare exceptions to this policy.

DEC 与其客户的关系则恰恰相反。PDP-1 是出售的，而不是出租的。DEC 不仅允许客户进行修改，还鼓励客户进行修改。PDP-1 的客户不多，但都很精明。第一个客户是剑桥咨询公司 Bolt Beranek and Newman (BBN)，该公司后来因其在创建互联网方面的作用而闻名。其他客户包括劳伦斯利弗莫尔实验室、加拿大原子能公司和电信巨头 ITT。56^事实上，BBN 的 Edward Fredkin 在第一台 PDP-1 安装到麻省理工学院后提出了许多改进建议。奥尔森将另一台 PDP-1 捐赠给了麻省理工学院，它成为了后来在民间传说中被推崇的黑客文化的基础，在那里成为了传奇。这些学生蜂拥而至，而不是等待将一副副牌提交给校园 IBM 大型机。其最著名的应用之一是作为技术模型铁路俱乐部布局的控制器。⁵⁷显然，大型计算机使用的经济学（不仅在商业设施中实行，而且在麻省理工学院自己的大型计算机设施中也实行）并不适用于 PDP-1。

The relationship DEC developed with its customers grew to be precisely the opposite. The PDP-1 was sold, not leased. DEC not only permitted, it encouraged modification by its customers. The PDP-1’s customers were few, but they were sophisticated. The first was the Cambridge consulting firm Bolt Beranek and Newman (BBN), which later became famous for its role in creating the Internet. Others included the Lawrence Livermore Laboratory, Atomic Energy of Canada, and the telecommunications giant, ITT.⁵⁶ Indeed, a number of improvements to the PDP-1 were suggested by Edward Fredkin of BBN after the first one was installed there. Olsen donated another PDP-1 to MIT, where it became legendary as the basis for the hacker culture later celebrated in popular folklore. These students flocked to the PDP-1 rather than wait their turn to submit decks of cards to the campus IBM mainframe. Among its most famous applications was as a controller for the Tech Model Railroad Club’s layout.⁵⁷ Clearly the economics of mainframe computer usage, as practiced not only at commercial installations but also at MIT’s own mainframe facility, did not apply to the PDP-1.

DEC很快开始发布有关其产品内部工作原理的详细规范，并广泛分发这些规范。肯尼斯·奥尔森的兄弟、该公司的一名员工斯坦·奥尔森说，他希望为 Digital 的产品提供一份“西尔斯罗巴克目录”之类的东西，其中包含大量关于如何将它们相互连接以及如何将它们连接到外部工业或实验室设备的教程信息。58^在斯坦的建议下，与业内其他公司的政策相反，DEC 将这些手册印在新闻纸上，装订得很便宜，每本只需几美分（图 4.2）。DEC 的销售员随身携带着这些手册，并将它们慷慨地分发给他们的客户或他们认为可能是客户的几乎任何人。

DEC soon began publishing detailed specifications about the inner workings of its products, and it distributed them widely. Stan Olsen, Kenneth Olsen’s brother and an employee of the company, said he wanted the equivalent of “a Sears Roebuck catalog” for Digital’s products, with plenty of tutorial information on how to hook them up to each other and to external industrial or laboratory equipment.⁵⁸ At Stan’s suggestion, and in contrast to the policy of other players in the industry, DEC printed these manuals on newsprint, cheaply bound and costing pennies a copy to produce (figure 4.2). DEC salesmen carried bundles of these around and distributed them liberally to their customers or to almost anyone they thought might be a customer.

鼓励客户了解和修改其产品的政策是出于必要。这家位于阿萨贝特磨坊一角的小公司无力开发将通用计算机变成有用产品所需的专用接口、安装硬件和软件。IBM 有能力做到这一点，但 DEC 别无选择，只能让客户了解其产品内部工作原理，而对于其他公司来说，这些秘密是他们小心翼翼保守的。令许多人惊讶的是，DEC 发现，客户不仅不介意这项工作，而且还欢迎这个机会^。59

This policy of encouraging its customers to learn about and modify its products was one borne of necessity. The tiny company, operating in a corner of the Assabet Mills, could not afford to develop the specialized interfaces, installation hardware, and software that were needed to turn a general-purpose computer into a useful product. IBM could afford to do that, but DEC had no choice but to let its customers in on what, for other companies, were jealously guarded secrets of the inner workings of its products. DEC found, to the surprise of many, that not only did the customers not mind the work but they welcomed the opportunity.⁵⁹

PDP-8 1965 年首次推出的产品是 PDP-8（图 4.3 ），它揭示了这个市场的规模。DEC 安装了超过 50,000 台 PDP-8 系统，加上多年后开发的不计其数的单芯片实现^。60

The PDP-8 The product that revealed the size of this market was one that was first shipped in 1965: the PDP-8 (figure 4.3). DEC installed over 50,000 PDP-8 systems, plus uncounted single-chip implementations developed years later.⁶⁰

PDP-8 的字长为 12 位，DEC 工程师将其起源追溯到与 Foxboro 公司就过程控制应用进行的讨论。他们还承认 12 位 CDC-160 对他们的决定产生了影响。61^另一个影响是林肯实验室的 Wes Clark 设计的一台名为 LINC 的计算机，这是一台 12 位机器，旨在供实验室环境中的人员用作个人计算机。62^在C. Gordon Bell 的领导下，在 Edson DeCastro 负责逻辑设计的情况下，DEC 于 1963 年底推出了一台 12 位计算机 PDP-5。两年后，他们推出了性能大大改进的继任产品 PDP-8。

The PDP-8 had a word length of 12 bits, and DEC engineers have traced its origins to discussions with the Foxboro Corporation for a process-control application. They also acknowledge the influence of the 12-bit CDC-160 on their decision.⁶¹ Another influence was a computer designed by Wes Clark of Lincoln Labs called the LINC, a 12-bit machine intended to be used as a personal computer by someone working in a laboratory setting.⁶² Under the leadership of C. Gordon Bell, and with Edson DeCastro responsible for the logic design, DEC came out with a 12-bit computer, the PDP-5, in late 1963. Two years later they introduced a much-improved successor, the PDP-8.

图 4.2

DEC 手册。DEC 将这些技术手册印在廉价的新闻纸上，并免费赠送给任何有兴趣使用小型计算机的人。（来源： Mark Avino，NASM。）

Figure 4.2

DEC manuals. DEC had these technical manuals printed on cheap newsprint, and the company gave them away free to anyone who had an interest in using a minicomputer. (Source : Mark Avino, NASM.)

PDP-8 的成功以及它所催生的小型计算机现象归功于性能、存储、封装和价格等多种因素的融合。性能是其中一个因素。PDP-8 的电路使用由“微合金扩散”工艺制成的锗晶体管，该工艺由 Philco 为其命运多舛的 S-2000 系列首创。这些晶体管的运行速度明显快于采用其他技术制成的晶体管。（PDP-8 每秒可以执行大约 35,000 次加法。）⁶³ 12 位字长严重限制了 PDP-8 可直接访问的内存量。一个字的 7 位组成地址字段；这样可以访问 2 ⁷或 128 个字。PDP-8 通过两种方式解决了这个限制。其一是使用“间接寻址”，在地址字段中指定一个内存位置，该位置不包含所需的数据，而是包含该数据的地址。（这样就可以使用一个字的全部 12 位而不是只有 7 位来作为地址。）另一种方法是将内存划分为单独寻址的“页面”，利用这样一个事实：大多数时候，人们只是从一小部分内存访问数据；只有偶尔计算机才需要跳转到另一页。这个过程并不像直接寻址内存那么简单，但如果这种情况发生得不太频繁，它也不会减慢速度。

The PDP-8’s success, and the minicomputer phenomenon it spawned, was due to a convergence of a number of factors, including performance, storage, packaging, and price. Performance was one factor. The PDP-8’s circuits used germanium transistors made by the “micro-alloy diffused” process, pioneered by Philco for its ill-fated S-2000 series. These transistors operated at significantly higher speeds than those made by other techniques. (A PDP-8 could perform about 35,000 additions per second.)⁶³ The 12-bit word length severely limited the amount of memory a PDP-8 could directly access. Seven bits of a word comprised the address field; that gave access to 2⁷ or 128 words. The PDP-8 got around that limitation in two ways. One was to use “indirect addressing,” to specify in the address field a memory location that contained not the desired piece of data but the address of that data. (This allowed for the full 12 bits of a word instead of only seven to be used for an address.) The other was to divide the memory into separately addressed “pages,” exploiting the fact that most of the time one is accessing data from a small portion of memory; only occassionally would the computer have to jump to another page. That process was not as simple as addressing memory directly, but it did not slow things down if it did not happen too often.

图 4.3

数字设备公司 PDP-8。计算机的逻辑模块安装在控制面板上方的两个塔上。通常，这些模块都用烟熏塑料封装。请注意左侧电路板上的分立电路：原始 PDP-8 使用分立电路，而不是集成电路。（来源： Laurie Minor，Smithsonian。）

Figure 4.3

Digital Equipment Corporation PDP-8. The computer’s logic modules were mounted on two towers rising from the control panel. Normally these were enclosed in smoked plastic. Note the discrete circuits on the boards on the left: The original PDP-8 used discrete, not integrated circuits. (Source : Laurie Minor, Smithsonian.)

逻辑和核心内存技术的改进将内存周期时间缩短到 1.6 微秒 — 比 IBM 7090 稍快，比 CDC 160 快四倍，比 20 世纪 50 年代末最快的鼓式计算机 Bendix G-15 快一千多倍。64 ^PDP -8 的字长较短，这意味着它在执行 10 位十进制数或浮点数的算术运算方面无法与其大型机竞争对手竞争，但对于许多其他应用而言，它的速度与任何价格的计算机一样快。65^这种性能使 PDP-8 及其之后的小型计算机与 G-15、LGP-30、IBM 1401 和其他“小型”计算机有着根本的不同。

Improvements in logic and core memory technology reduced the memory cycle time to 1.6 microseconds—slightly faster than the IBM 7090, four times faster than the CDC 160, and over a thousand times faster than the Bendix G-15, the fastest drum computer of the late 1950s.⁶⁴ The PDP-8’s short word length meant that it could not compete with its mainframe competitors in doing arithmetic on 10-digit decimal or floating-point numbers, but for many other applications it was as fast as any computer one could buy at any price.⁶⁵ That kind of performance made the PDP-8 and the minicomputers that followed it fundamentally different from the G-15, the LGP-30, the IBM 1401, and other “small” computers.

基本 PDP-8 带有 4000 字的内存，分为 32 个块，每个块 128 个字。通过设置指令字操作码中的两位之一，可以跨块或“页面”进行访问。对于外部内存，DEC 提供了一种简单、廉价但功能强大的磁带系统，该系统源自 LINC。他们称之为“DECtape”。与大型机磁带系统相比，一卷 DECtape 轻便易携；驱动器结构紧凑，可以与计算机本身放在同一个设备机架中。数据可以以 128 个字的块为单位双向读取或写入，而不仅仅是附加在记录末尾。DECtape 的作用更像是现代个人计算机上的软盘驱动器，而不是大型机磁带驱动器的档案存储方式^。66

The basic PDP-8 came with four thousand words of memory, divided into 32 blocks of 128 words each. Access across a block, or “page,” was possible by setting one of two bits in the operation code of an instruction word. For external memory DEC provided a simple, inexpensive, but capable tape system derived from the LINC. They called it “DECtape.” Again in contrast to mainframe tape systems, a reel of DECtape was light and portable; the drive was compact and could fit into the same equipment rack as the computer itself. Data could be read or written in either direction, in blocks of 128 words, not just appended at the end of a record. DECtape acted more like the floppy disk drives on modern personal computers, than like the archival storage style of mainframe tape drives.⁶⁶

PDP-8 的物理封装对大型系统来说不那么重要，但它的成功却起了关键作用。PDP-8 使用了一系列紧凑的模块，上面安装了晶体管、电阻器和其他组件。每个模块执行明确定义的逻辑功能（类似于第一个集成电路执行的功能）。这些模块又被插入一个像书一样打开的铰链底盘中。结果是一个由处理器、控制面板和核心内存组成的系统，封装小到可以嵌入到其他设备中。模块本身通过绕线互连（参见第 2 章）。DEC 使用 Gardner-Denver 公司的自动绕线机为 PDP-8 接线。这消除了接线错误，使 DEC 能够处理它很快收到的大量订单。这台计算机占用 8 立方英尺的体积，重 250 磅^。67

The physical packaging of the PDP-8, a factor that mattered less for large systems, played a key role in its success. The PDP-8 used a series of compact modules, on which transistors, resistors, and other components were mounted. Each module performed a well-defined logic function (similar to the functions that the first integrated circuits performed). These in turn were plugged into a hinged chassis that opened like a book. The result was a system consisting of processor, control panel, and core memory in a package small enough to be embedded into other equipment. The modules themselves were interconnected by wire-wrap (see chapter 2). DEC used automatic wire-wrapping machinery from the Gardner-Denver Corporation to wire the PDP-8. This eliminated wiring errors and allowed DEC to handle the large orders it soon received. The computer occupied eight cubic feet of volume and weighed 250 pounds.⁶⁷

PDP-8 的定价问题。低价可以促进销售，但也可能使 DEC 无法获得足够的收入来支持研发，而研发是 DEC 保持技术领先地位所必需的（避免 20 世纪 50 年代中期许多新兴计算机公司的命运，这些公司最终被 Burroughs 或 NCR 等老牌公司收购）。DEC 的高管决定冒这个险，他们将 PDP-8 定价为 18,000 美元，包括用于 I/O 的电传打字机终端。几年之内，一台 PDP-8 的价格不到 10,000 美元。低价震惊了计算机行业，并引发了大批订单。再一次，对计算机市场规模的所有估计都显得过于谨慎。68^包括IBM 在内的老牌公司最终进入了这个市场，但 DEC 继续发展壮大。它首先找到了一种方法，即通过继续借鉴麻省理工学院研究界的知识和技能，保持计算机技术的领先地位。它还继续保持较低的运营成本。设在一家旧毛纺厂当然有帮助，但更重要的是 DEC 与客户建立的关系，客户负责开发工作和相关成本。（这将在稍后讨论。）

There was the matter of pricing the PDP-8. A low price would generate sales, but it might also prevent DEC from generating enough revenue to support research and development, which it would need to keep its lead in technology and (avoid the fate of many of the start-up computer companies of the mid-1950s, which ended up being bought by established companies like Burroughs or NCR). Executives at DEC decided to take the risk, and they priced the PDP-8 at $18,000, including a teletype terminal for I/O. Within a few years one could be bought for less than $10,000. The low price shocked the computer industry and generated a flood of orders. Once again all estimates of the size of the market for computers turned out to be too timid.⁶⁸ Established companies, including IBM, eventually entered this market, but DEC continued to grow and prosper. It found a way, first of all, to stay at the forefront of computer technology by continuing to draw from the knowledege and skills of the MIT research community. It also continued to keep the cost of its operations low. Being based in an old woolen mill certainly helped, but even more important was the relationship DEC developed with its customers, who took responsibility for development work and associated costs. (This will be discussed shortly.)

PDP-8 使用了电传打字机公司的一种新设备——33 型 ASR（“自动发送/接收”）来加载和编辑程序。69^它比早期小型计算机使用的 Flexowriter 更便宜、更简单、更耐用（图 4.4）。与 Flexowriter 一样，它也可以用作打字机，可以打印到一卷连续的纸上、将指示按下了哪个键的代码直接发送到计算机或将该代码打到纸带上。数据传输速率为每秒 6 到 10 个字符。33 型于 20 世纪 60 年代中期推出，是首批采用当时由美国标准协会颁布的位编码标准的计算机之一，该标准称为 ASCII（美国信息交换标准代码）。Flexowriter 的代码受到一些商业设备公司的欢迎，但在 ASCII 开发出来后，它的代码被拒绝作为计算机行业的基础。⁷⁰就像链式打印机象征着大型计算机环境一样，33 型打印机象征着小型计算机时代以及随后的个人计算机时代的开始。它对个人计算产生了深远的影响，尤其是在键盘方面：例如，控制键和退出键首次出现在 33 型打印机上。十五年后，该机器特有的许多其他键代码进入了个人计算机软件，但很少有人知道它们是如何进入的。

For loading and editing programs the PDP-8 used a new device from the Teletype Corporation, the Model 33 ASR (“automatic sendreceive”).⁶⁹ It was cheaper, simpler, and more rugged than the Flexowriter used by earlier small computers (figure 4.4). Like the Flexowriter, it functioned as a typewriter that could print onto a roll of continuous paper, send a code indicating what key was pressed directly to a computer, or punch that code onto a paper tape. Data were transmitted at a rate from six to ten characters per second. Introduced in the mid-1960s, the Model 33 was one of the first to adopt the standard for coding bits then being promulgated by the American Standards Association, a code known as ASCII (American Standard Code for Information Interchange). The Flexowriter’s code was popular with some business equipment companies, but its code was rejected as a basis for the computer industry when ASCII was developed.⁷⁰ Just as the Chain Printer symbolized the mainframe computing environment, the Model 33 came to symbolize the minicomputer era and the beginnings of the personal computer era that followed it. It had a far-reaching effect on personal computing, especially on the keyboard: the control and escape keys, for example, first made their general appearance on the Model 33. Many other key codes peculiar to this machine found their way into personal computer software fifteen years later, with few people realizing how they got there.

最后，还有计算机的名称。“Minicomputer”很吸引人，符合时代潮流，并赋予了 PDP-8 独特的个性。人们可以购买一台小型计算机，而不必签订严格的租赁协议、拥有恒温室或拥有一支似乎会将用户拒之门外的技术团队。迷你裙恰好在 PDP-8 开始销售时出现（来自英国），毫无疑问，它的一些魅力被转移到了计算机上。可能是 DEC 驻欧洲的销售员为 PDP-8 起了这个名字。⁷¹（鉴于 Kenneth Olsen 的保守宗教背景，他不太可能想出这个名字。作为斯堪的纳维亚人后裔，他既不抽烟，也不喝酒，更不会说脏话。）这个名字的另一个来源，与 PDP-8 完美契合，也是英国出口的——Morris Mini-Minor，由传奇汽车工程师 Alec Issigonis 设计，以应对 1956 年苏伊士运河危机导致的波斯湾石油供应中断。Issigonis 的设计重量轻、反应灵敏、运行经济。最重要的是，它的表现优于与其竞争的大多数笨重、臃肿的英国汽车。英国向世界各地出口 Mini-Minor 和迷你裙。数字设备公司在小型计算机方面也做了同样的事情。

Finally, there was the computer’s name. “Minicomputer” was catchy, it fit the times, and it gave the PDP-8 an identity. One could obtain a minicomputer and not feel obliged also to get a restrictive lease agreement, a climate-controlled room, or a team of technicians whose job seemed to be keeping users away. The miniskirt happened to come along (from Britain) at the time the PDP-8 was beginning to sell, and no doubt some of its glamour was transferred to the computer. It may have been a DEC salesman stationed in Europe who gave the PDP-8 that name. ⁷¹ (Given Kenneth Olsen’s conservative religious upbringing, it was unlikely that he would have come up with it. Of Scandinavian descent, he neither smoked nor drank nor used profanity.) Another source of the name, one that fits the PDP-8 perfectly, was also a British export—the Morris Mini-Minor, designed by the legendary automobile engineer Alec Issigonis, in response to the Suez Canal Crisis that cut off Persian Gulf oil to Britain in 1956. Issigonis’s design was lightweight, responsive, and economical to operate. Most important, it outperformed most of the stodgy, bloated British cars with which it competed. The British exported Mini-Minors and miniskirts around the world. Digital Equipment Corporation did the same with minicomputers.

图 4.4

ASR-33 电传打字机，早期小型计算机的标准输入/输出设备，尽管它最初并非为此目的而设计的。请注意“Control”（CTRL）和“Escape”（ESC）键，它们后来成为台式计算机键盘的标准。“X-ON”（CTRL-Q）和“X-OFF”（CTRL-S）命令也嵌入到个人计算机操作系统中。“@”符号（Shift-P）后来被用来指示互联网上的地址。（来源：明尼苏达大学查尔斯·巴贝奇研究所。）

Figure 4.4

An ASR-33 Teletype, the standard input/output device for early minicomputers, although it was not originally designed for that purpose. Note the “Control” (CTRL) and “Escape” (ESC) keys, which later became standard for desktop computer keyboards. The “X-ON” (CTRL-Q) and “X-OFF” (CTRL-S) commands also became embedded into personal computer operating systems. The “@” symbol (Shift-P) was later adopted for indicating addresses on the Internet. (Source : Charles Babbage Institute, University of Minnesota.)

编写 PDP-8 程序来做一些有用的事情需要相当的技巧。由于内存有限，程序员们放弃了高级编程语言，转而使用汇编语言甚至机器代码。但是，PDP-8 架构的简单性，加上 DEC 的免费信息政策，使它成为一台易于理解的计算机。这些因素的结合催生了所谓的原始设备制造商 (OEM)；一家独立的公司购买小型计算机，添加用于输入和输出的专用硬件，为最终系统编写专用软件，并以自己的品牌出售（高价出售）。“OEM”一词的起源不明。在一些早期的参考文献中，它暗示计算机制造商，而不是第三方，是 OEM，这似乎是“原始设备”的合理定义。然而，最终，这一含义完全与围绕小型计算机构建系统的一方有关^。72

Programming a PDP-8 to do something useful required no small amount of skill. Its limited memory steered programmers away from high-level programming languages and toward assembly or even machine code. But the simplicity of the PDP-8’s architecture, coupled with DEC’s policy of making information about it freely available, made it an easy computer to understand. This combination of factors gave rise to the so-called original equipment manufacturer (OEM); a separate company that bought minicomputers, added specialized hardware for input and output, wrote specialized software for the resulting systems, and sold them (at a high markup) under its own label. The origin of the term “OEM” is obscure. In some early references it implies that the computer manufacturer, not the third party, is the OEM, which seems a logical definition of “original equipment.” Eventually, however, the meaning attached entirely to the party that built systems around the mini.⁷²

与 OEM 打交道使小型计算机制造商无需开发专门的软件。DEC 开发了一些自己的应用程序，比如计算机排版系统，但那是例外。73^俄勒冈州希尔斯伯勒的 Electronics Diversified 公司生产的 LS-8 是一种典型的 OEM 产品，它用于操作剧院舞台灯光，通过编程序列控制一组灯光。LS-8 的功能被认为是百老汇长期热播剧《歌舞线上》成功的关键因素。74 LS-8 里面是一台 PDP-8A，这是 DEC 于 1975 年推出的型号。LS-8 的用户不一定知道这一点，因为 LS-8 有自己的控制面板，不是为计算机用户量身定制的，而是^为剧院灯光工作人员量身定制的。OEM 应用涵盖社会各阶层，从医疗器械到小型企业记录保存，再到工业控制器。一套基于 PDP-8 的系统甚至被安装在马铃薯采摘机上，并放在拖拉机的后面（图 4.5）^。75

Dealing with an OEM relieved the minicomputer manufacturer of the need to develop specialized software. DEC developed some applications of its own, such as the computerized typesetting system, but that was the exception.⁷³ A typical OEM product was the LS-8 from Electronics Diversified of Hillsboro, Oregon, which it was used to operate theatrical stage lighting, controlling a complex of lights through programmed sequences. The LS-8’s abilities were cited as a key element in the success of the long-running Broadway hit A Chorus Line.⁷⁴ Inside the LS-8 was a PDP-8A, a model that DEC had introduced in 1975. Users of the LS-8 did not necessarily know that, because the LS-8 had its own control panel, tailored not to computer users but to theatrical lighting crews. OEM applications ranged across all segments of society, from medical instrumentation to small business record keeping, to industrial controllers. One PDP-8–based system was even installed in a potato-picking machine and carried on the back of a tractor (figure 4.5).⁷⁵

DEC 文化Alec Issigonis 认为 Morris Mini-Minor 成功的关键在于它是由一个不超过 6 人的能干工程团队设计的，管理层允许该团队在很少或根本没有外界干扰的情况下进行工作。76^这是对数字设备公司文化最贴切的描述。77^尽管发展迅速，DEC 仍然保留着小公司的氛围，产品开发的责任落在小团队的工程师身上。1965 年，DEC 的营收为 1500 万美元，拥有 876 名员工。到 1970 年，DEC 的营收为 1.35 亿美元，拥有 5800 名员工。78^这只是 IBM 规模的一小部分，尽管 DEC 的 PDP-8 计算机出货量与 IBM 360 系列出货量相当。

The DEC Culture Alec Issigonis believed that the key to the success of the Morris Mini-Minor was that it was designed by a capable engineering team of no more than six persons, which was permitted by management to operate with little or no outside interference.⁷⁶ That is about as good a description of the culture at Digital Equipment as one could hope to find.⁷⁷ Though growing fast, DEC retained the atmosphere of a small company where responsibility for product development fell to small groups of engineers. In 1965 it had revenues of $15 million and 876 employees. By 1970 DEC had revenues of $135 million and 5,800 employees.⁷⁸ That was a small fraction of IBM’s size, although DEC was shipping as many PDP-8 computers as IBM was shipping of its 360 line.

图 4.5

安装在拖拉机上并控制土豆采摘机的 PDP-8。尽管安装起来有些不方便，但它预示着微处理器将嵌入农场和其他地方几乎所有复杂机械的时代。（来源：数字设备公司。）

Figure 4.5

A PDP-8 mounted on a tractor and controlling a potato-picker. Although an awkward installation, it foreshadowed the day when microprocessors were embedded into nearly all complex machinery, on the farm and elsewhere. (Source : Digital Equipment Corporation.)

随着 Digital 公司逐渐成为 IBM 的主要竞争对手之一，它依然保持着斯巴达风格——甚至过于斯巴达了。Digital 公司逐渐接管了 Assabet 工厂的越来越多的部分，直到最终将其全部买下（图 4.6）。在工厂中寻找出路是一件令人生畏的事情，但在那里工作的“工厂老鼠”记住了走廊、桥梁和通道的位置。Digital 公司在邻近的城镇开设了分支机构，但“工厂”仍然是该公司的精神中心。客户不断惊叹于它的简洁和朴实。一位华尔街分析师毫不掩饰地轻蔑地表示，该公司在搬进来之前只是“勉强翻新”了这座 19 世纪的工厂。79^退伍军人管理局的一位管理人员正在改装 DEC 设备以在手术期间监测大脑功能，他表达了类似的惊讶：

As Digital grew into one of IBM’s major competitors, it remained Spartan—excessively so. Digital gradually took over more and more of the Assabet Mills, until it eventually bought it all (figure 4.6). Finding one’s way through the complex was daunting, but the “Mill rats” who worked there memorized the location of the corridors, bridges, and passageways. Digital opened branch facilities in neighboring towns, but “the Mill” remained the spiritual center of the company. Customers were continually amazed at its simplicity and lack of pretension. One Wall Street analyst said, with unconcealed scorn, that the company had only “barely refurbished” the nineteenth-century mill before moving in.⁷⁹ An administrator from the Veterans Administration, who was adapting DEC equipment for monitoring brain functions during surgery, expressed similar surprise:

我不知道你是否去过原来的工厂，但它是（或曾经是）一个漂亮的 19 世纪老工厂，在内战期间用于制造羊毛毯，所以木地板上浸满了羊毛脂，偶尔需要擦拭。这是一座巨大的建筑，晚上没有人的时候在里面工作有点吓人^。80

I don’t know if you’ve ever been to the original factory, but it is (or was) a nice old nineteenth-century mill that was used to make wool blankets during the civil war, so the wooden floors were soaked with lanolin and had to be swabbed occasionally. It was a huge building, and a little spooky to work in at night when no one else was around.⁸⁰

图 4.6

马萨诸塞州梅纳德的 The Mill。数字设备公司总部。（来源：数字设备公司。）

Figure 4.6

The Mill, Maynard, Massachusetts. Headquarters for Digital Equipment Corporation. (Source : Digital Equipment Corporation.)

一位来自美国中西部一所小型学院的英语教授，想使用 PDP-8 对 17 和 18 世纪伦敦舞台上的数据进行排序和分类，他这样描述了他第一次参观 Mill 的感受：

A professor of English from a small midwestern college, who wanted to use a PDP-8 to sort and classify data on the London Stage in the seventeenth and eighteenth centuries, described his first visit to the Mill this way:

梅纳德仍保留着足够的乡村气息，让人想起梭罗曾经在树林里漫步。像许多新英格兰城镇一样，梅纳德中心正上方的河流上建有一座水坝，水坝下方是一堆堆老旧的红砖工厂，颜色逐渐变淡为紫色。DEC 似乎占据了梅纳德中心的所有工厂建筑，它们都通过桥墩以某种角度连接在一起，河水以某种方式穿过它们。

Maynard is still rural enough to remind one that Thoreau once roamed its woods. Like many New England towns it has a dam in its river just above the center and a jumble of old red brick mills mellowing toward purple beneath the dam. DEC apparently occupied all the mill buildings in Maynard Center, and they were all connected by abutment at some angle or another by covered bridges, and the river got through them somehow.

从游客的破旧沥青停车场到主入口，是一条横跨沟渠的木桥，通往一栋工厂大楼的上层。进入阁楼后，进入一个相当宽敞、灯光明亮、朴素、没有地毯的区域，阁楼的尽头有一个柜台和一扇门。柜台上一位慈母般的女士帮您将来事记在卡片上，并请您在房间中间一排大约七把椅子上坐下。那里有几本破旧的杂志可以看。无法推断出他们选择杂志的原则，也无法推断出他们来到这里是一系列偶然事件。科罗拉多州市政当局，《猫爱好者文摘》，《今日心理学^》。81

The main entrance from the visitors’ disintegrating asphalt parking lot was a wooden footbridge across a gully into an upper floor of one of the factory buildings. One entered a fairly large, brightly lighted, unadorned, carpetless section of a loft with a counter and a door at the far end. At the counter a motherly person helped one write down one’s business on a card and asked one to take a seat in a row of about seven chairs down the middle of the room. There were a few dog-eared magazines to look at. It was impossible to deduce the principle of their selection or the series of accidents by which they had arrived here. Colorado Municipalities, Cat-Lover’s Digest, Psychology Today.⁸¹

人们对 Digital 产生了狂热的迷恋，许多客户，尤其是科学家或工程师同行，被其斯巴达式的形象所鼓舞。DEC 代表了计算机的一切解放，而 IBM 的着装规范和最重要的打孔卡代表了一切错误。82^华尔街分析师习惯于企业财富和权力的象征，他们认为密尔文化表明该公司不是像 IBM 或 UNIVAC 那样的严肃的计算机公司。83^更重要的是，DEC 的营销策略（包括向销售人员支付薪水而不是佣金）非常低。一些人认为情况比这更糟：DEC “蔑视”营销，因此错失了比现在更大发展的机会。84 ^DEC的增长速度不如同时成立的另一家公司 Control Data 或 Scientific Data Systems，但它的 PDP-8 销售速度与生产速度一样快，并且它正在开拓 CDC 和 SDS 都未进入的新计算机市场。正是这最后一种品质让该公司与众不同。从 20 世纪 90 年代的角度来看，人们可以说 DEC 只是一家成长、繁荣然后被时代所淹没的计算机公司。但这样说会忽略一个事实，即 DEC 将计算重新定位为我们现在认为的“自然”或显而易见的计算定义方式。如果不了解计算对 Assabet Mills 工程师的影响，就不可能理解 20 世纪末计算的状况。

A cult fascination with Digital arose, and many customers, especially scientists or fellow engineers, were encouraged to buy by the Spartan image. DEC represented everything that was liberating about computers, while IBM, with its dress code and above all its punched card, represented everything that had gone wrong.⁸² Wall Street analysts, accustomed to the trappings of corporate wealth and power, took the Mill culture as a sign that the company was not a serious computer company, like IBM or UNIVAC.⁸³ More to the point, DEC’s marketing strategy (including paying their salesmen a salary instead of commissions) was minimal. Some argued it was worse than that: that DEC had “contempt” for marketing, and thus was missing chances to grow even bigger than it did.⁸⁴ DEC did not grow as fast as Control Data or Scientific Data Systems, another company that started up at the same time, but it was selling PDP-8s as fast as it could make them, and it was opening up new markets for computers that neither CDC nor SDS had penetrated. It was this last quality that set the company apart. One could say from the perspective of the 1990s that DEC was just another computer company that grew, prospered, and then was eclipsed by events. But that would miss the fact that DEC reoriented computing toward what we now assume is the “natural” or obvious way to define computing. It is impossible to understand the state of computing at the end of the twentieth century without understanding computing’s debt to the engineers at the Assabet Mills.

但无论 DEC 的形象如何，它都不认为自己是一家只制造小型计算机的公司。在推出 PDP-8 的同时，DEC 还推出了一款大型系统，即 36 位 PDP-6。这款计算机只售出了 23 台，但其改进版 PDP-10 却成为许多大学计算机科学系和其他高端客户的最爱。PDP-10 于 1966 年首次推出，从一开始就被设计为支持分时和传统批处理。然而，除了少数有影响力的使用者之外，PDP-10 对 IBM 凭借其 7090 和 360 系列机器占据主导地位的大型机业务的影响很小。

But whatever its image, DEC did not see itself as a company that built only small computers. Simultaneously with the PDP-8 it introduced a large system, the 36-bit PDP-6. Only twenty-three were sold, but an improved version, the PDP-10, became a favorite of many university computer science departments and other sophisticated customers. First delivered in 1966, the PDP-10 was designed from the start to support time-sharing as well as traditional batch processing. Outside the small though influential group of people who used it, however, the PDP-10 made only a small dent on the mainframe business that IBM dominated with its 7090 and 360-series machines.

从 20 世纪 70 年代末开始，DEC 凭借其 VAX 产品线最终成为大型系统市场的有力竞争者。到那时，它也已经消除了 Mill 文化中的一些缺陷。其销售人员继续领取薪水，但在其他方面，DEC 销售人员与 IBM 类似。Digital 仍留在 Mill，但翻新了访客接待区，使其与其他大型公司类似。（然而，由于它位于 Maynard 中部，停车位仍然有限；访客只需将车停在市中心的街道上，小心地将几枚硬币放入计价器中，以免被罚款。Maynard 仍然是一个节俭的新英格兰小镇。）砖墙仍然在那里，装饰着几件精心挑选的织布机或梳棉机零件，这些零件是羊毛厂时代遗留下来的。访客可以向坐在设备齐全的安全台旁的接待员报上自己的名字，坐在舒适而现代的椅子上，在等待预约时阅读《华尔街日报》。到 20 世纪 80 年代末，制造厂已迁往海外或马萨诸塞州和新罕布什尔州各地分布的更现代化、更实用的建筑。工厂现在成了坐在办公桌前的办公室工作人员的办公场所，而不是坐在工作台上的工程师的办公场所。奥尔森的继任者罗伯特·B·帕尔默 (Robert B. Palmer) 于 1993 年决定将公司总部迁出工厂，迁往梅纳德一栋较小的现代化建筑。大约在同一时间，有消息称公司将更名为 Digital，而不是 DEC——这是一个小小的变化，但在某种程度上象征着一个时代的逝去。小型计算机时代结束了，但这只是在它彻底改变了计算之后。

DEC did eventually became a serious contender in the large systems market with its VAX line, beginning in the late 1970s. By that time it had also smoothed the rougher edges off of the Mill culture. Its sales force continued to draw a salary, but in other respects DEC salesmen resembled IBM’s. Digital remained in the Mill but refurbished the visitors’ reception area so it resembled that of any other large corporation. (Because of its location in the middle of Maynard, however, there still was limited parking; visitors simply parked on a downtown street, being careful to put a few dimes into the meter to keep from getting a ticket. Maynard still was a thrifty New England town.) The brick walls were still there, adorned with a few well-chosen pieces of a loom or carding machine leftover from the woolen mill days. A visitor could announce his or her name to a receptionist seated at a well-appointed security desk, settle into a comfortable and modern chair, and peruse the Wall Street Journal while waiting for an appointment. By the late 1980s the manufacturing had moved overseas or to more modern and utilitarian buildings scattered throughout Massachusetts and New Hampshire. The Mill was now a place for office workers seated at desks, not for engineers at workbenches. Olsen’s successor, Robert B. Palmer, decided in 1993 to move the company’s headquarters out of the Mill and into a smaller, modern building in Maynard. Around the same time word went out that the company was to be called Digital, not DEC—a small change but somehow symbolic of the passing of an age. The era of the minicomputer came to an end, but only after it had transformed computing.

与 MIT 的联系Mill 是了解 DEC 进入计算业务方式的一个线索。在该公司 1992 年出版的公司历史中可以找到一个更具启发性的线索（当时个人计算机正在对 DEC 的业务构成挑战）。《数字化工作》的第一章讨论的不是 Mill、PDP-1 或 Olsen，而是“MIT 和旋风传统”。85^本章以一张 MIT 主楼的照片开篇。书中第一张人物照片是 MIT 学生；接下来是教授和旋风项目工作人员（Jay Forrester、Robert Everett 和 JA O'Brien）的照片。

The MIT Connection The Mill was one clue to DEC’s approach to entering the computing business. A more revealing clue is found in a corporate history that the company published in 1992 (when the personal computer was challenging DEC’s business). The first chapter of Digital at Work is a discussion not of the Mill, the PDP-1, or of Olsen, but of “MIT and the Whirlwind Tradition.”⁸⁵ The chapter opens with a photograph of MIT’s main building. The first photographs in the book of people are of MIT students; next are photos of professors and of the staff (Jay Forrester, Robert Everett, and J. A. O’Brien) of Project Whirlwind.

Whirlwind 计算机于 1950 年开始运行，到 DEC 成立时，它已经过时了。但是，Whirlwind 项目奠定的基础足以支撑 DEC 多年后的发展。Whirlwind 最引人注目的衍生品是 SAGE 防空系统。DEC、小型计算机以及在波士顿郊区兴起的其他计算机公司是其他更重要的衍生品。Ken Olsen 与 Georges Doriot 联手，找到了一种方法，将 MIT 的工程研究氛围（其最伟大的倡导者是 Jay Forrester）带出校园，远离军事资助，带入一家商业公司。这种方法非常巧妙，而且被重复使用了很多次，事后看来，这是理所当然和显而易见的。尽管也有平行转移到私营部门，但二战和冷战早期武器实验室（雷达、核裂变、超音速空气动力学、弹道导弹）的其他产品很少有这样的轨迹。计算能力，而不是核能，已经变得“便宜到无法计量”。

The Whirlwind computer was operational in 1950, and by the time DEC was founded it was obsolete. But the foundations laid by Project Whirlwind were stong enough to support DEC years later. The most visible descendant of Whirlwind was the SAGE air-defense system. DEC, the minicomputer, and the other computer companies that sprouted in suburban Boston were other, more important offspring. Ken Olsen, allied with Georges Doriot, found a way to carry the MIT atmosphere of engineering research, whose greatest exponent was Jay Forrester, off the campus, away from military funding, and into a commercial company. It was so skillfully done, and it has been repeated so often, that in hindsight it appears natural and obvious. Although there have been parallel transfers to the private sector, few other products of World War II and early Cold War weapons labs (radar, nuclear fission, supersonic aerodynamics, ballistic missiles) have enjoyed this trajectory. Computing, not nuclear power, has become “too cheap to meter.”

许多人认为，这种新的技术创业文化是美国 20 世纪 90 年代经济繁荣的主要推动力，这种文化比密尔的氛围持续的时间更长。它被成功地移植到了西海岸的硅谷（尽管由于尚不清楚的原因，波士顿周围的 128 号公路，后来被称为技术高速公路，逐渐衰落）。在硅谷，斯坦福大学和伯克利大学取代了麻省理工学院，国防高级研究计划局 (DARPA) 取代了美国海军和空军。86^旧金山出现了许多仿效多里奥特的美国研究与发展公司的风险投资公司。许多分析这一现象的畅销书都忽略了它的大学根源；另一些则未能理解军事资助的作用。有些书专注于硅谷百万富翁的财富和奢侈的生活方式——几乎不适用于肯·奥尔森，他朴素的生活是传奇。

That new culture of technical entrepreneurship, considered by many to be the main force behind the United States’s economic prosperity of the 1990s, lasted longer than the ambience of the Mill. It was successfully transplanted to Silicon Valley on the West Coast (although for reasons yet to be understood, Route 128 around Boston, later dubbed the Technology Highway, faded). In Silicon Valley, Stanford and Berkeley took the place of MIT, and the Defense Advanced Research Projects Agency (DARPA) took over from the U.S. Navy and the Air Force.⁸⁶ A host of venture capital firms emerged in San Francisco that were patterned after Doriot’s American Research and Development Corporation. Many of the popular books that analyze this phenomenon miss its university roots; others fail to understand the role of military funding. Some concentrate on the wealth and extravagant lifestyles adopted by the millionaires of Silicon Valley—hardly applicable to Ken Olsen, whose plain living was legendary.

IBM 代表了约翰·肯尼斯·加尔布雷斯所说的“技术结构”的完美体现：一个大型、高度组织化、垂直整合的公司，它控制、管理并引导技术创新的混乱局面，从而占据市场主导地位。IBM 平稳运营的核心是当时畅销书《组织人》（作者威廉·怀特）中的一个人物。87^人们取笑这位穿着白衬衫和保守西装、严格遵循“IBM 方式”的 IBM 员工。然而，他们当中谁不嫉妒公司的利润和 IBM 销售人员赚取的丰厚佣金？仔细阅读怀特的书，就会发现他由衷钦佩这样的人，没有他们，公司几乎无法生存，更不用说繁荣了。奥尔森利用了另一个知识来源；他别无选择。奥尔森和他刚从麻省理工学院毕业的年轻工程师们也是“组织人”，只是类型不同。他们也拥有一套共同的价值观，只是他们的价值观来自麻省理工学院校园里的旧临时建筑，战争期间辐射实验室就设在那里。这些价值观似乎与 IBM 的价值观截然不同，但它们足以将 DEC 员工塑造成一个具有竞争力的组织。这些工程师驳斥了当时的智慧，即孤独先驱的时代已经结束，初创公司永远无法与巨头竞争。

IBM represented the perfection of what John Kenneth Galbraith called the “technostructure”: a large, highly organized, vertically integrated firm that controlled, managed, and channeled the chaos of technical innovation into market dominance. Central to smooth operations at IBM was a character from a best-seller from that era, The Organization Man, by William Whyte.⁸⁷ People made fun of the IBM employee, with his white shirt and conservative suit, who followed the “IBM way” so closely. Yet who among them was not jealous of the company’s profits and the generous commissions earned by IBM salesmen? A closer reading of Whyte’s book reveals a genuine admiration for such people, without whom a company could hardly survive, let alone prosper. Olsen tapped into an alternate source of knowledge; he had no choice. Olsen and his young engineers just out of MIT were “organization men,” too, only of a different stripe. They, too, shared a set of common values, only theirs came from the old temporary buildings on the MIT campus, the ones where the Radiation Lab was housed during the War. Those values seemed very different from IBM’s, but they were strong enough to mold DEC employees into a competitive organization. These engineers refuted the wisdom of the day, which stated that the era of the lone pioneer was over, that start-up companies could never compete against the giants.

PDP-8 的朴素外观掩盖了它所引发的巨大影响。大型计算机将继续存在，尽管其统治时代已经屈指可数。只要经济条件允许，许多人将继续通过打牌来使用计算机。IBM 将继续主导该行业。计算机业务不是零和游戏；DEC 的收益并不一定就是 IBM 的损失——至少在一段时间内不是。小型计算机表明，只要有合适的包装、价格，以及最重要的是，用户能够更直接地使用计算机，就会打开全新的市场。这无异于重新定义“计算机”一词，其重要性不亚于 20 世纪 40 年代的重新定义，当时这个词的意思是机器，而不是进行计算的人。要发挥这一潜力，还需要二十年的技术发展。最终，数字设备公司以及 IBM 和其他大型机公司将受到阿萨贝特工厂所释放的力量的冲击，而事实证明，这种力量是无法抑制的。

The modest appearance of the PDP-8 concealed the magnitude of the forces it set into motion. Mainframe computing would persist, although its days of domination were numbered. As long as the economics were in its favor, many would continue to use a computer by punching decks of cards. IBM would continue to dominate the industry. The computer business was not a zero-sum game; DEC’s gain was not automatically IBM’s loss—at least not for a while. The mini showed that with the right packaging, price, and above all, a more direct way for users to gain access to computers, whole new markets would open up. That amounted to nothing less than a redefinition of the word “computer,” just as important as the one in the 1940s, when that word came to mean a machine instead of a person that did calculations. Fulfilling that potential required two more decades of technical development. Ultimately Digital Equipment Corporation, as well as IBM and the other mainframe companies, would be buffeted by the forces unleashed in the Assabet Mills, forces that would prove impossible to restrain.

“Go-Go”年代和 System/360，1961-1975

The “Go-Go” Years and the System/360, 1961–1975

IBM、七个小矮人和 BUNCH

IBM, the Seven Dwarfs, and the BUNCH

随着小型计算机在 20 世纪 60 年代中期确立了市场地位，大部分计算机资金仍用于购买 IBM 和少数竞争对手销售的大型主机。IBM 占据了商用市场约 70% 的份额，1963 年的收入为 12 亿美元，1965 年增长到 30 多亿美元，到 1970 年增长到 75 亿美元。1^仅次于IBM 的是 Sperry Rand，他是 20 世纪 40 年代原始 UNIVAC 和 ERA 开发的继承者，收入为 1.45 亿美元。美国市场的其他参与者包括 Control Data、Honeywell、Philco、Burroughs、RCA、General Electric 和 NCR。（AT&T 也生产计算机，但作为受监管的垄断企业，其数据在这里无法比较。）²

As the minicomputer established its markets in the mid-1960s, most computer dollars continued to be spent on large mainframes sold by IBM and a few competitors. IBM held about a 70% share of the commercial market, with 1963 revenues of $1.2 billion, growing to over $3 billion in 1965, and $7.5 billion by 1970.¹ Second to IBM was Sperry Rand, inheritor of the original UNIVAC and ERA developments of the 1940s, with $145 million in revenue. Other players in the U.S. market were Control Data, Honeywell, Philco, Burroughs, RCA, General Electric, and NCR. (AT&T also manufactured computers, but as a regulated monopoly its figures are not comparable here.)²

除了 Control Data 之外，上述所有公司都专注于 IBM 所倡导的相同计算模式：大型集中式主机安装，运行以打孔卡形式提交的批量程序。3^那些希望参与该业务竞争的公司从底层到顶层提供一切——硬件、外围设备、系统和应用软件以及服务。他们试图通过提供计算机租赁和直接销售来与 IBM 进一步竞争。这需要大量资金，而 IBM 以外的所有公司的利润都很低或没有利润。

With the partial exception of Control Data, all the above companies focused on the same model of computing espoused by IBM: large, centralized mainframe installations, running batches of programs submitted as decks of punched cards.³ Those who wished to compete in this business provided everything from bottom to top—hardware, peripherals, system and applications software, and service. They sought further to compete with IBM by offering to lease as well as sell their computers outright. That required enormous amounts of capital, and profits for everyone except IBM were low or nonexistent.

当时参与者的地位让 IBM 专家称他们为“白雪公主和七个小矮人”。这个称呼颇具讽刺意味：“白雪公主”经常成为“七个小矮人”（例如 Control Data）或联邦政府本身的诉讼对象，理由是他们存在垄断行为。到 20 世纪 70 年代，通用电气和 RCA 退出了该行业，IBM 的竞争对手因此有了一个新的名称“BUNCH”（Burroughs、UNIVAC、NCR、Control Data 和 Honeywell）。这个组合一直稳定到 20 世纪 80 年代 — 在计算机这样一个动荡的行业中，这一点尤其突出。20 世纪 80 年代个人计算机的出现改变了整个行业的性质，主机供应商的简单分组被瓦解。

The status of the players at the time led IBM-ologists to call them “Snow White and the Seven Dwarfs.” The term was ironic: “Snow White” was periodically the target of lawsuits either from one of the “Dwarfs” (e.g., Control Data) or the Federal government itself, for monopoly practices. By the 1970s General Electric and RCA had left the business, leading to a new term for IBM’s competitors, the “BUNCH” (Burroughs, UNIVAC, NCR, Control Data, and Honeywell). This constellation remained stable into the 1980s—remarkably so in an industry as volatile as computers. The advent of personal computers in the 1980s changed the nature of the entire business, and the simple grouping of mainframe suppliers unraveled.

IBM 系统/360

IBM System/360

当 DEC 于 1965 年初开始发售其 PDP-8 时，IBM 交付了一系列大型机中的第一台，这些大型机将使该公司在业界占据了更有利的地位。这就是 1964 年 4 月发布的 System/360（图 5.1）。之所以这样命名，是因为它面向从商业到科学的各个领域，既面向进行大量数学计算的客户，也面向对大量数据进行简单算术运算的客户。System/360 的主要卖点是 IBM 提供的不是一种而是一整套计算机，并承诺为一种型号编写的程序可以运行在更大的型号上，从而随着业务的增长节省客户的软件投资。IBM 于 1964 年 4 月 7 日发布了六种型号。后来，它又发布了其他型号，但在开始交付时放弃了原有六种型号中的一些。这个想法并不完全是新的：计算机公司在推出新型号时曾试图保持软件兼容性，就像 IBM 对其 704、709 和 7090 机器所做的那样。但 360 是一系列计算机，它们同时发布，性能范围约为 25:1。除了 20 世纪 50 年代末交付给陆军的一小部分机器外，以前从未有人尝试过^。4

As DEC began shipping its PDP-8 in early 1965, IBM delivered the first of a series of mainframes that would propel that company into an even more commanding position in the industry. That was the System/360, announced in April 1964 (figure 5.1). It was so named because it was aimed at the full circle of customers, from business to science, at customers who did a lot of mathematical calculation and at those who did simpler arithmetic on large sets of data. System/360’s primary selling point was that IBM was offering not one but a whole line of computers, with a promise that programs written for one model would work on larger models, thus saving a customer’s investment in software as business grew. IBM announced six models on April 7, 1964. Later on it announced others, while dropping some of the original six by the time deliveries began. The idea was not entirely new: computer companies had tried to preserve software compatibility as they introduced newer models, as IBM had done with its 704, 709, and 7090 machines. But the 360 was a series of computers, all announced at the same time, offering about a 25 : 1 performance range. Except for a small run of machines delivered to the Army in the late 1950s, that had never been attempted before.⁴

图 5.1

IBM System/360。IBM 的宣传照片，展示了 1964 年发布的产品的巨大规模和范围（来源： IBM。）

Figure 5.1

IBM System/360. A publicity photo from IBM, showing the vast size and scope of products announced in 1964 (Source : IBM.)

有一句经常被重复的短语，最早出现在《财富》杂志的一篇文章中，IBM 的一名员工说“你把公司押在了”这一系列电脑上。5^除了这六种电脑型号外，IBM 还在同一天推出了“150 多种不同的产品——新磁带、新磁盘、029 打卡机”。6^如果360 失败了，那将是一次毁灭性的打击，尽管 IBM 仍将作为该行业的主要参与者生存下来。该公司本可以推出其备受推崇的 1401 和 7090 系列机器的更新版本，并且仍然有来自计算机出现之前的打卡机安装的稳定收入来源。但这样的失败将重组计算机行业^。7

In an often-repeated phrase, first used in a Fortune magazine article, an IBM employee said “you bet your company” on this line of computers.⁵ Besides the six computer models, IBM introduced “over 150 different things—new tapes, new disks, the 029 card punch” on the same day.⁶ Had the 360 failed, it would have been a devastating blow, although IBM would still have survived as a major player in the business. The company could have introduced newer versions of its venerable 1401 and 7090-series machines, and it still had a steady stream of revenue from precomputer punched card installations. But such a failure would have restructured the computer industry.⁷

System/360 没有失败。1964 年 4 月产品发布后几周内，订单就开始纷至沓来。“System/360 计算机的订单很快就超出了预期：第一个月就收到了 1100 多台。五个月后，数量翻了一番，相当于美国安装的 IBM 计算机数量的五分之一” ⁸基本架构成为 IBM 产品线在 20 世纪 90 年代的支柱。

System/360 did not fail. Within weeks of the product announcement in April 1964 orders began coming in. “Orders for System/360 computers promptly exceeded forecasts: over 1100 were received in the first month. After five months the quantity had doubled, making it equal to a fifth of the number of IBM computers installed in the U.S.”⁸ The basic architecture served as the anchor for IBM’s product line into the 1990s.

生产和交付计算机产品线需要大量资源。公司扩大了生产设施，但交付时间表却落后了，并且出现了关键部件短缺的情况。360 的成功对公司的生存威胁几乎不亚于失败。对于那些被逼到绝境的员工来说——这样的员工很多——IBM 收入的增长可能不值得他们承受身体和精神上的压力。从 1965 年到 1970 年，主要得益于 System/360，IBM 的总收入增长了一倍多。净收益也翻了一番，到 1971 年超过了 10 亿美元。自 20 世纪 50 年代中期以来，IBM 一直引领着美国计算机行业。到 1970 年，它的计算机安装基数已达到 35,000 台，到 20 世纪 70 年代中期，可以将美国计算机行业描述为两个相等的部分：一边是 IBM，另一边是其他所有公司的总和^。9

Manufacturing and delivering the line of computers required enormous resources. The company expanded its production facilities, but delivery schedules slipped, and shortages of key components arose. The success of the 360 threatened the company’s existence almost as much as a failure might have. For those employees driven to the breaking point— and there were many—the jump in revenues for IBM may not have been worth the physical and mental stress. From 1965 to 1970, thanks mostly to System/360, IBM’s gross income more than doubled. Net earnings also doubled, surpassing $1 billion by 1971. IBM had led the U.S. computer industry since the mid 1950s. By 1970 it had an installed base of 35,000 computers, and by the mid-1970s it made sense to describe the U.S. computer industry as having two equal parts: IBM on one side and everyone else combined on the other.⁹

IBM 在努力满足需求时面临的问题（员工倦怠、错过发货日期、生产线质量控制）可能是其竞争对手所希望遇到的问题。显然，许多客户发现这一系列机器很合他们的心意。例如，大多数 NASA 中心迅速从 7090 型设备切换到 360 型（65 型或更高型号），以满足将人类送上月球的需求。使用计算机进行商业数据处理的商业公司同样用 System/360 型号替换了 7030 型和其他系统。用低端 360 替换古老的 1401 型遇到了一些阻力，但总体而言，市场对适用于科学和商业应用的兼容机器系列的想法给予了压倒性的认可（图 5.2）。

The problems IBM faced in trying to meet the demand—employee burnout, missed shipping dates, quality control on the production lines—were problems its competitors might have wished for. Obviously many customers found this line of machines to their liking. Most NASA centers, for example, quickly switched over to 360 (Model 65 or higher) from their 7090 installations to meet the demands of putting a man on the Moon. Commercial firms that used computers for business data processing likewise replaced their 7030s and other systems with models of the System/360. There was some resistance to replacing the venerable 1401 with the low-end 360, but in general the marketplace gave overwhelming approval to the notion of a compatible family of machines suitable for scientific as well as business applications (figure 5.2).

促成 System/360 诞生的决策来自 IBM 的一个委员会，即 SPREAD。该委员会于 1961 年末在康涅狄格州科斯科布的 Sheraton New Englander 汽车旅馆召开了两个月的每日会议。他们的报告于 1961 年 12 月 28 日内部发布，并在 22 年后发表，揭示了当时计算领域的许多现状，以及 IBM 的主要工程师和高管对计算领域的设想^。10

The decisions that led to System/360 came from an IBM committee known as SPREAD, which met daily in the Sheraton New Englander motel in Cos Cob, Connecticut, for two months in late 1961. Their report, issued internally on December 28, 1961, and published twenty-two years later, reveals much about the state of computing, as it then existed, and as key engineers and executives at IBM thought it would become.¹⁰

图 5.2

小型 System/360 安装。请注意背景中的真空柱式磁带驱动器和前面带有 Selectric 机制的打字机。最前景是磁盘驱动器。（来源： IBM 档案。）

Figure 5.2

A small-scale System/360 installation. Note the vacuum-column tape drives in the background and a typewriter with the Selectric mechanism in the front. In the extreme foreground is a disk drive. (Source : IBM Archives.)

他们的讨论始于对公司现有产品的调查。1961 年，IBM 推出了一堆令人困惑的机器，其中很少有兼容的。其中两台脱颖而出，已经进行了描述。1401 是一款小型字符型计算机，以低廉的价格出租，深受人们的喜爱。它面向商业客户，也因随附的 1403 型链式打印机而广受科学界欢迎。1401 的销量达数千台。另一台出色的机器是 7090/7094 科学计算机。它价格昂贵，但性能出色，深受 NASA 和大型航空航天公司等客户的欢迎。它的销量只有几百台。IBM 的大型商用计算机 7070 的销售情况令人失望，而小型科学计算机 1620 的销售情况良好，尽管不如^1401。11

Their deliberations began with a survey of the company’s existing products. In 1961 IBM was fielding a confusing tangle of machines, few of which were compatible with one another. Two of them stood out and have already been described. The 1401 was a small, character-oriented computer that rented at a low price and was well liked. Intended for business customers, it was also popular for scientific use, mainly due to the excellent Model 1403 chain printer that came with it. Sales of the 1401 were measured in the thousands. The other outstanding machine was the 7090/7094 scientific computer. It was expensive, but its performance made it popular with customers like NASA and the large aerospace firms. Its sales measured in the low hundreds. IBM’s large business computer, the 7070, had had disappointing sales, while the small scientific machine, the 1620, was doing well, although not as well as the 1401.¹¹

所有这些机器都是晶体管机器，不过 7090 是真空管 709 的晶体管版本，为了赢得空军合同而匆忙推出。由于 7090 和 1401 都是在旧基础上构建的，因此很难实现性能的数量级提升。而且它们彼此不兼容。12^与此同时，商业和科学用途之间的区别观念已经站不住脚。根据这一观念，商业客户处理大量数据并对其进行简单算术运算，而科学客户则相反，对小型数据集进行高级计算。然而事实上，除了需要浮点运算之外，科学家和工程师还需要在有限元分析等应用中处理大量数据集，有限元分析是一种为构建复杂的航空航天结构而开发的技术。13 随着^联邦医疗保险等计划在工作场所的普及，工资单等常规商业交易的复杂性也日益增加。 SPREAD 委员会由来自 IBM 两条产品线的成员组成，他们最初并不同意统一产品线，但最终他们认识到了它的优势，并将其作为建议纳入最终报告中。与许多伟大的想法一样，现在回想起来，统一产品线的概念似乎显而易见，但对于聚集在汽车旅馆房间里的人来说，情况并非如此^。14

All were transistorized machines, although the 7090 was a transistorized version of the vacuum-tube 709, hastily introduced to win an Air Force contract. Because both the 7090 and the 1401 were built on an old foundation, it would have been difficult to achieve an order-of-magnitude increase in performance for either. And they were incompatible with each other.¹² Meanwhile, the notion of what divided business from scientific use was not holding up. According to that notion, business customers handled large sets of data, on which they performed simple arithmetic, while scientific customers did the opposite, advanced calculation on small sets of data. In fact, however, along with a need for floating-point arithmetic, scientists and engineers also needed to handle large data sets in applications like finite-element analysis, for instance, a technique developed for building complex aerospace structures.¹³ And routine business transactions like payroll had increasing complexity, as federal programs like Medicare spread through the workplace. The SPREAD Committee, composed of members from both of IBM’s product lines, did not agree at first on a unified product line, but eventually they recognized its advantages and incorporated that as a recommendation in their final report. As with many great ideas, the notion of having a unified product line seems obvious in retrospect, but that was not the way it seemed at first to those assembled in the rooms of the motel.¹⁴

可扩展性则不那么明显。尽管 SPREAD 委员会同意这是必要的，但在早期阶段，弗雷德·布鲁克斯和吉恩·阿姆达尔（后来成为 360 的主要设计者）都认为“这不可能”。15^很少有其他技术可以简单地扩展。例如，土木工程师在设计大坝时使用的标准与设计小坝时使用的标准不同。大型轿车的发动机、变速箱、动力传动系和车架不仅仅是为超小型车设计的发动机、变速箱、动力传动系和车架的放大版。SPREAD 委员会提议的计算范围是 25:1——更像是将超小型车与 18 轮卡车进行比较。然而，到 1970 年，在 IBM 宣布升级 360 系列后，它提供了具有 200:1 范围的兼容计算机^。16

Less obvious was scalability. Even though the SPREAD Committee agreed that this was needed, at the early stages both Fred Brooks and Gene Amdahl—later two of the 360’s principal architects—argued that “it couldn’t be done.”¹⁵ Few other technologies, if any, scale simply. Civil engineers, for example, use different criteria when designing large dams than they use for small ones. The engine, transmission, power train, and frame of a large sedan are not simply bigger versions of those designed for a subcompact. What the SPREAD Committee was proposing was a range of 25 : 1 in computing—more like comparing a subcompact to an 18-wheeler. By 1970, however, after IBM had announced an upgrade to the 360 line, it was offering compatible computers with a 200 : 1 range.¹⁶

改变布鲁克斯和阿姆达尔想法的，是重新发现了一个几乎和存储程序计算机一样古老的概念。1951 年，在曼彻斯特大学数字计算机落成典礼的演讲中，莫里斯威尔克斯提出“设计自动计算机的最佳方式”是将其控制部分构建为一个独立的小型存储程序计算机，其中每个控制操作（比如，将两个数字相加的命令）被分解为一系列“微操作”，由存储“微程序”的组件矩阵指挥。17^通过增加设计的复杂性，威尔克斯实际上简化了它。控制单元的设计通常是最困难的，现在可以由一系列更简单的电路组成，就像计算机的存储单元一样。18^威尔克斯大胆断言这是“最佳方式”，因为他认为这会让设计更合乎逻辑、更规则、更简单；他几乎是事后才想到，“直到机器建造的后期才需要最终决定订单代码。” ¹⁹他并没有说任何关于一系列机器或计算机具有一定功率范围的事情。

What changed Brooks’s and Amdahl’s mind was the rediscovery of a concept almost as old as the stored-program computer itself. In 1951, at a lecture given at a ceremony inaugurating the Manchester University digital computer, Maurice Wilkes argued that “the best way to design an automatic calculating machine” was to build its control section as a little stored-program computer of its own, wherein each control operation (say, the command to add two numbers) is broken down into a series of “micro-operations” directed by a matrix of components that stored a “micro-programme [sic].”¹⁷ By adding a layer of complexity to the design, Wilkes in fact simplified it. The design of the control unit, typically the most difficult, could now be made up of an array of simpler circuits, like those for the computer’s memory unit.¹⁸ Wilkes made the bold assertion that this was the “best way” because he felt it would give the design more logical regularity and simplicity; almost as an afterthought he mentioned that “the order code need not be decided on finally until a late stage in the construction of the machine.”¹⁹ He did not say anything about a series of machines or computers having a range of power.

这个想法在后来的曼彻斯特活动中得到了延续，SPREAD 委员会成员 John Fairclough 在那里学习电气工程。他提出了使用微编程（采用美式拼写）作为在 360 系列中实施一组通用指令的方法的想法，同时允许负责每种特定型号详细设计的工程师优化设计，以降低成本并实现足够的性能。微程序以小型只读存储器的形式内置于每种型号处理器的控制单元中，将编写以确保兼容性。微编程使 360 的设计人员“能够将设计过程……与有效体现我们试图开发的机器的指令集特征的控制逻辑分开。” ²⁰

The idea was kept alive in later activity at Manchester, where John Fairclough, a member of the SPREAD Committee, studied electrical engineering. Through him came the notion of using microprogramming (adopting the American spelling) as a way of implementing a common set of instructions across the line of 360s, while allowing the engineers charged with the detailed design of each specific model to optimize the design for low cost and adequate performance. The microprogram, in the form of a small read-only memory built into the control unit of each model’s processor, would be written to ensure compatibility. Microprogramming gave the 360’s designers “the ability to separate the design process ... from the control logic that effectively embodied the instruction-set characteristics of the machine we were trying to develop.”²⁰

IBM 对这一概念的采纳扩展了威尔克斯的原始见解。从本质上讲，这是对通用存储程序计算机基本属性的重新表述——通过在一个层面上接受复杂性（计算机需要大量组件），人们在另一个层面上获得了强大功能和简单性（组件采用规则数组的形式，可以通过数学和逻辑工具进行分析）。在最早的数字机器中，对这一概念的一些理解似乎还处于萌芽阶段。威尔克斯本人可能受到了贝尔实验室继电器计算机 Model VI 的启发，他可能在 1950 年访问美国期间检查过该计算机。在 Model VI 上，一组线圈存储了信息，使机器能够在从纸带接收到一条简单指令后执行复杂的子序列^。21

IBM’s adoption of this concept extended Wilkes’s original insight. In essence it is a restatement of the fundamental property of a general-purpose, stored-program computer—that by accepting complexity at one level (computers require very large numbers of components), one gains power and simplicity at another level (the components are in the form of regular arrays that can be analyzed by tools of mathematics and logic). Some understanding of this concept appears inchoate in the earliest of the digital machines. Wilkes himself may have been inspired by the Bell Labs relay computer Model VI, which he probably inspected during a visit to America in 1950. On the Model VI a set of coils of wire stored information that allowed the machine to execute complex subsequences upon receiving one simple instruction from a paper tape.²¹

^{通过采用微编程，IBM 获得了另一个优势，有些人认为这}是 360 最初成功的关键。22那就是能够安装微程序，使处理器能够理解为早期 IBM 计算机编写的指令。通过这种方式，IBM 销售人员可以说服客户采用新技术，而不必担心突然导致应用软件投资过时。IBM 的 Larry Moss 将这种能力称为仿真，暗示它“与原版一样好”（甚至更好），而不仅仅是“模拟”或更糟的“模仿”。360 Model 65 的销量特别好，因为它能够模拟大型商用计算机 7070，而 IBM 则为低端型号 30 和 40 投入了额外的资源来模拟^1401。23

By adopting microprogramming, IBM gained one further advantage, which some regard as the key to the 360’s initial success.²² That was the ability to install a microprogram that would allow the processor to understand instructions written for an earlier IBM computer. In this way IBM salesmen could convince a customer to go with the new technology without fear of suddenly rendering an investment in applications software obsolete. Larry Moss of IBM called this ability emulation, implying that it was “as good as” (or even better than) the original, rather than mere “simulation” or worse, “imitation.” The 360 Model 65 sold especially well because of its ability to emulate the large business computer 7070, and IBM devoted extra resources to the low-end models 30 and 40 to emulate the 1401.²³

理论上，任何存储程序计算机都可以通过编程来像另一台计算机一样运行——这是“通用图灵机”的结果。图灵机以数学家艾伦·M·图灵的名字命名，他于 20 世纪 30 年代提出了这一概念。在实践中，这通常意味着不可接受的性能损失，因为额外的代码层会减慢速度。试图用一台计算机模拟另一台计算机通常会让倒霉的设计者陷入“图灵焦油坑”：一切皆有可能，但没有什么是切合实际的。24 ³⁶⁰避开了这个坑，因为它的模拟使用了软件和每台机器控制单元的微程序的组合（图 5.3）。当与它也使用的更快的电路相结合时，这种组合使新机器运行旧程序的速度比在 1401 上运行相同程序的速度快10 倍。根据估计，到 1967 年，超过一半的 360 应用程序中都是对旧硬件的模拟。

In theory, any stored-program computer can be programmed to act as if it were another—a consequence of its being a “Universal Turing Machine,” named after the mathematician Alan M. Turing, who developed this concept in the 1930s. In practice, that usually implies an unacceptable loss of performance, as the extra layers of code slow things down. Trying to emulate one computer with another usually lands the hapless designer in the “Turing Tar-Pit,” where anything is possible but nothing is practical.²⁴ The 360 avoided that pit because its emulation used a combination of software and the microprogram of each machine’s control unit (figure 5.3). When combined with the faster circuits it also used, the combination permitted the new machines to run the old programs as much as ten times faster than the same program would have run on, say, a 1401. By 1967, according to some estimates, over half of all 360 applications were emulations of older hardware.

图 5.3

IBM 9020 配置控制台。9020 由一组三台 System/360、Model 50 计算机组成，配置为实时运行。该系统设计为在任何一台或两台单台计算机发生故障时都能正常运行。直到 1997 年夏天，9020 系统一直用于美国各地的民航空中交通管制作业。（来源： Terry McCrae，Smithsonian。）

Figure 5.3

IBM 9020 Configuration Control Console. The 9020 consisted of a set of three System/360, Model 50 computers, configured to operate in real time. The system was designed to run correctly during the failure of any one or even two individual computers. The 9020 systems were used for en route civilian air traffic control operations throughout the United States until the summer of 1997. (Source : Terry McCrae, Smithsonian.)

1401 仿真对于 IBM 孤注一掷的赌注来说尤其重要：1963 年 12 月，霍尼韦尔推出了 H-200 计算机，他们称之为“Liberator”的程序允许它运行 1401 程序。H-200 的销售立即火爆起来，而 IBM 当时宣布 360 系列与 1401 不兼容。销售 1401 的 IBM 部门在 1964 年初陷入了低迷，但在 360 低端型号的订单滚滚而来后，他们又走出了困境。25^仿真的成功证明了计算机术语的一个悖论：尽管软件的名字如此，但它比硬件更永久且更难修改。直到今天，来自各种供应商的现代计算机上仍有 1401 个程序在运行常规工资单和其他数据处理工作。 20 世纪 60 年代初，程序员们使用打字机编写这些工作时，并不知道他们的工作能持续多久。（1401 软件的寿命是“2000 年”漏洞的主要原因。）

1401 emulation was especially crucial to IBM’s bet-the-company gamble: In December 1963 Honeywell introduced the H-200 computer, with a program they called “Liberator” that allowed it to run 1401 programs. H-200 sales were immediately brisk, just as IBM was announcing the 360 line with its implied incompatibility with the 1401. The IBM division that sold the 1401 went through a Slough of Despond in early 1964, but it climbed out after orders for the lower-end models of the 360 came rolling in.²⁵ The success of emulation demonstrated a paradox of computer terminology: software, despite its name, is more permanent and hard to modify than hardware. To this day there are 1401 programs running routine payroll and other data-processing jobs, on modern computers from a variety of suppliers. When programmers coded these jobs in the early 1960s using keypunch machines, they had no idea how long-lived their work would be. (The longevity of 1401 software was a major cause of the “Year-2000” bug.)

System/360 还有其他值得一提的架构特性。其中许多特性最初是在一个名为 STRETCH 的系统中引入的，该系统是为洛斯阿拉莫斯国家实验室设计的，于 20 世纪 60 年代初完成。这个名字源于它的目标，即“扩展”最先进的处理速度。在 1961 年向洛斯阿拉莫斯交付 STRETCH 后，IBM 推出了一个商业版本，即 IBM 7030，但在交付了 8 次之后，它撤回了它并注销了巨额财务损失。后来 IBM 意识到这也许并不是失败，因为许多最初在 STRETCH 中探索的概念都进入了 System/ ^360。26

The System/360 had other architectural features worth mentioning. Many of these were first introduced in a system called STRETCH, designed for the Los Alamos National Laboratory and completed in the early 1960s. The name came from its goal, to “stretch” the state of the art in processing speed. After delivering the STRETCH to Los Alamos in 1961, IBM marketed a commercial version as the IBM 7030, but after eight deliveries it withdrew it and wrote off a large financial loss. Later on IBM realized that perhaps it was not a failure after all, since so many concepts first explored in STRETCH found their way into the System/360.²⁶

除了最小的 Model 20 之外，每个 System/360 的中央处理器都包含 16 个通用寄存器。几乎所有以前的计算机设计都指定了一个寄存器，即累加器，用于进行简单的算术和逻辑运算；另一个寄存器，即索引寄存器或“B 行”，用于保存用于内存访问的索引信息。其他寄存器可能专用于其他特殊功能。在 360 中，这 16 个寄存器中的任何一个都可以用于任何操作（除了少数例外，例如用于浮点数的额外寄存器）。

Every System/360 except for the smallest Model 20 contained sixteen general-purpose registers in its central processor. Nearly all previous computer designs specified one register, the accumulator, where simple arithmetic and logical operations took place; another register, the index register or “B-line,” held indexing information for memory access. Still other registers might be dedicated to other special functions. In the 360, any of the sixteen registers could be used for any operation (with a few exceptions, like extra registers for floating-point numbers).

360 的字长为 32 位——比 7090/7094 科学计算机的字长短 4 位，但由于 32 是 2 的幂，因此简化了设计。大多数早期计算机使用 6 位来编码字符；IBM System/360 使用 8 位，IBM 的 Werner Buchholz 在 1956 年将其称为“字节”。27^因为8 也是 2 的幂，这进一步简化了机器的逻辑。它还允许每个字符有 28^或256 种不同的组合；这对于大写和小写字母、十进制数字 1 到 10、标点符号、重音符号等来说已经足够了。而且由于 4 位足以编码一个十进制数字，因此可以将两个十进制数字“打包”到每个字节中，而 6 位字节中只能有一个十进制数字。（360 的内存是在字节级别寻址的；无法获取从字节中间开始的位序列。）

The 360’s word length was 32 bits—4 bits shorter than word length of the 7090/7094 scientific computers, but because 32 was a power of 2, it simplified the design. Most early computers used sets of 6 bits to encode characters; System/360 IBM used 8 bits, which Werner Buchholz of IBM called a “byte,” in 1956.²⁷ Because eight is also a power of 2, this further simplified the machine’s logic. It also allowed 2⁸ or 256 different combinations for each character; which was more than adequate for upper-and lowercase letters, the decimal digits 1 to 10, punctuation, accent marks, and so on. And since 4 bits were adequate to encode a single decimal digit, one could “pack” two decimal digits into each byte, compared to only one decimal digit in a 6-bit byte. (The 360’s memory was addressed at the byte level; one could not fetch a sequence of bits that began in the middle of a byte.)

为了对这 256 种不同的组合进行编码，IBM 选择了他们为打孔卡设备开发的代码的扩展。这种扩展二进制编码的十进制交换码 (EBCDIC) 设计精良、功能齐全，并为未来的扩展提供了空间。但它有一个缺点 — 与同时开发的 ASCII 标准不兼容。ASCII 由美国国家标准协会于 1963 年支持，仅标准化了 7 位，而不是 8 位。原因之一是打孔纸带仍在广泛使用，委员会认为在标准纸带上打 8 个孔会使其过于脆弱。（还有其他一些原因。）由于缺乏 8 位标准，它不如 EBCDIC，但由于其官方地位，除 IBM 外，其他地方都采用了 ASCII。使用 ASCII 和电传打字机的小型计算机的快速普及进一步促进了该代码的传播。由于 IBM 在大型机安装方面占据主导地位，这两种标准都无法胜过另一种。²⁸ IBM 曾在开发 ASCII 的委员会中派有代表，System/360 也规定可以使用任一代码，但后来 ASCII 模式被放弃，因为它很少使用。²⁹几年之内采用两个不兼容的标准令人遗憾，但可能并不奇怪。后来还会发生类似事件。³⁰

To encode the 256 different combinations, IBM chose an extension of a code they had developed for punched card equipment. This Extended Binary Coded Decimal Interchange Code (EBCDIC) was well designed, complete, and offered room for future expansion. It had one unfortunate characteristic—incompatibility with the ASCII standard being developed at the same time. ASCII, supported by the American National Standards Institute in 1963, standardized only seven bits, not eight. One reason was that punched paper tape was still in common use, and the committee felt that punching eight holes across a standard piece of tape would weaken it too much. (There were a few other reasons as well.) The lack of an 8-bit standard made it inferior to EBCDIC, but because of its official status, ASCII was adopted everywhere but at IBM. The rapid spread of minicomputers using ASCII and Teletypes further helped spread the code. With the dominance by IBM of mainframe installations, neither standard was able to prevail over the other.²⁸ IBM had had representatives on the committee that developed ASCII, and the System/360 had a provision to use either code, but the ASCII mode was later dropped as it was little used.²⁹ The adoption of two incompatible standards within a few years of each other was unfortunate but probably not surprising. Similar events would occur later on.³⁰

这两个标准的传播只产生了一些影响。在 ASCII 中，十个十进制数字的编码值低于字母表中的字母；而在 EBCDIC 中则相反。因此，如果数据以 ASCII 编码，排序程序会将“3240”排在“Charles”之前，但如果使用 EBCDIC，则将“Charles”排在“3240”之前。在 EBCDIC 中，可能反映了其穿孔卡的起源，字节的八位从左到右按顺序编号，最左边的位代表最高有效位；例如，代表 2 ⁸个值的位是位 #0，2 ⁷是位 1，依此类推。在 ASCII 中则相反。从工程角度来看，这是一个微不足道的差异，大多数用户永远不必担心它。31由于在小型计算机中的领先地位， ^ASCII从 20 世纪 80 年代开始在个人计算机和工作站环境中占据主导地位。

There have been only a few consequences of the spread of these two standards. In ASCII, the ten decimal digits were encoded with lower numerical values than the letters of the alphabet; with EBCDIC it was the opposite. Therefore a sorting program would sort “3240” before “Charles” if the data were encoded in ASCII, but “Charles” before “3240” if EBCDIC had been used. In EBCDIC, possibly reflecting its punched card ancestry, the eight bits of a byte were numbered sequentially from left to right, with the leftmost representing the most significant bit; for example, the bit representing the 2⁸th value was bit #0, 2⁷ bit 1, and so on. With ASCII it was the reverse. From an engineering standpoint this is a trivial difference, and most users never have to worry about it.³¹ Because of its beachhead in minicomputers, ASCII would prevail in the personal computer and workstation environment beginning in the 1980s.

360 的设计者允许用 4 位字寻址 16 个通用寄存器，用 24 位寻址机器的核心内存。这样可以直接访问 2 ^{24 ，即 1600 万个地址，这在当时似乎}^足够了。与几乎所有其他计算机设计一样，地址空间最终被发现不足，1981 年 IBM 将地址位数扩展为 31，允许访问 20 亿个地址。32

The 360’s designers allowed for 4 bits of a word to address the 16 general-purpose registers, and 24 bits to address the machine’s core memory. That allowed direct access to 2²⁴, or 16 million addresses, which seemed adequate at the time. Like nearly every other computer design, the address space was eventually found to be inadequate, and in 1981 IBM extended the number of address bits to 31, allowing for access to 2 billion addresses.³²

对于较便宜的型号，即使允许 24 位也是奢侈的，因为它们旨在使用小得多的内存空间来完成工作。携带额外的地址位会带来系统开销损失，这可能会让霍尼韦尔等竞争对手提供更具成本效益的机器。IBM 的解决方案是在一条指令中只携带 24 个地址位中的 12 个。然后，将此数字添加到存储在“基址”地址寄存器中的另一个数字，以提供完整的 24 位地址。33^如果程序需要的内存少于 2 ¹²或 4 千字节，则无需转到基址寄存器。许多较小的问题都是这种情况，尤其是安装了较便宜的 360 型号的问题。对于较长的问题，转到基址寄存器获取地址当然会产生额外的开销损失，但实际上这并不是一个严重的问题。

For the cheaper models, even allowing 24 bits was extravagant, as these were intended to do their work with a much smaller memory space. Carrying the extra address bits would impose an overhead penalty that might allow competitors like Honeywell to offer machines that were more cost-effective. IBM’s solution was to carry only 12 of the possible 24 address bits in an instruction. This number would then be added to another number stored in a “base” address register to give the full 24-bit address.³³ If a program required fewer than 2¹² or four thousand bytes of memory, going to the base register was not necessary. That was the case for many smaller problems, especially those that the cheaper models of the 360 were installed for. For longer problems there was of course the additional penalty incurred when going to the base register to obtain an address, but in practice this was not a severe problem.

最后，System/360 保留了通过通道处理输入和输出的概念。有了标准接口，IBM 就可以提供适用于整个机器系列的单一磁带、卡和打印设备系列——这是一个强大的卖点，其优势很容易抵消为提供兼容性而必须做出的任何妥协。行业媒体将 I/O 设备称为“外围设备”，但它们是 System/360 项目的核心——新型键盘打孔机、新型磁盘和磁带驱动器，甚至带有著名高尔夫球打印头和经典键盘布局的 Selectric 打字机。所有这些设备都与米色倾斜控制面板一样定义了大型机计算的 360 时代^。34

Finally, the System/360 retained the concept of having channels to handle input and output. With a standard interface, IBM could offer a single line of tape, card, and printing equipment that worked across the whole line of machines—a powerful selling point whose advantages easily offset whatever compromises had to be made to provide compatibility. The trade press called I/O devices “peripherals,” but they were central to the System/360 project—a new model keypunch, new disk and tape drives, and even the Selectric typewriter with its famous golf-ball print head and classic keyboard layout. All of these devices defined the 360-era of mainframe computing as much as the beige, slanted control panel.³⁴

360 的架构设计采用了富有创意且有时非常巧妙的折衷方案，以实现跨性能范围的兼容性。最初它的设计相当简单，但多年来它变得越来越复杂、怪异和笨重。它能够像以前一样增长，足以在 20 世纪 90 年代保持活力，这一事实证明了最初努力的实力。

The architectural design of the 360 used creative and sometimes brilliant compromises to achieve compatibility across a range of performance. Initially it had a fairly simple design, but over the years it grew ever more complex, baroque, and cumbersome. The fact that it could grow as it did, enough to remain viable into the 1990s, is testimony to the strength of the initial effort.

System/360 和完整的计算周期 360 型号发布后不久，订单就开始源源不断地涌入，这证明了推出如此雄心勃勃、统一的产品线的决定是正确的。但这些订单是否掩盖了“360 度”中机器未能很好覆盖的任何领域？确实如此，尽管方式并不明显。

System/360 and the Full Circle of Computing The orders that began streaming in for models of the 360 shortly after it was announced validated the decision to offer such an ambitious, unified line of products. But did those orders mask any sectors in the “360 degrees” that the machines did not cover well? They did, though not in obvious ways.

第 4 章讨论了小型计算机的兴起，其中 PDP-8 是其中的佼佼者，它刚好是在 360 开始交付时推出的。随着小型计算机功能的增长，它们开始与 IBM 的大型机产品线竞争，但最初几乎没有重叠或竞争。PDP-8 不适用于大型机正在处理的工作，例如处理大量工资单，而 System/360 不适合用作实验室实验、实时数据收集和其他 PDP-8 特别擅长的用途的控制器。对于这些应用，IBM 提供了不兼容的硬件产品线。

Chapter 4 discussed the rise of the minicomputer, led by the PDP-8, which was introduced just as deliveries of the 360 began. As minicomputers grew in capability, they began to compete with IBM’s mainframe line, but initially there was little overlap or competition. The PDP-8 was not intended for the jobs the mainframes were being used for, such as processing large payrolls, and the System/360 was ill-suited as a controller for laboratory experiments, real-time data collection, and other uses that the PDP-8 was especially good for. For those applications, IBM offered an incompatible line of hardware.

分时和 System/360

Time-Sharing and System/360

然而，System/360 并未涵盖一个非常重要的领域——以交互方式或“对话方式”使用大型计算机。出于经济原因，无法将大型计算机专用于单个用户，因此实际上以交互方式使用大型计算机的唯一方法是让多个用户同时共享其计算周期或“时间”。

There was, however, one very important sector that System/360 did not cover—using a large computer interactively or “conversationally.” For economic reasons one could not dedicate a mainframe to a single user, so in practical terms the only way to use a large machine interactively was for several users to share its computational cycles, or “time,” simultaneously.

计算机的“分时”一词有多种含义。最早的含义相当有限，仅指程序员无需准备和提交一副新牌到系统作业队列中即可调试程序的能力。计算机执行一批作业所花费的时间将由程序员共享，程序员可以对程序进行一些更改，然后将其插入机器上已有的原始程序中。到 20 世纪 60 年代末，该术语有了更通用的定义，当时麻省理工学院的约翰·麦卡锡教授特别提倡这个定义，并将在以下对 System/360 的讨论中使用。根据该定义，每个用户都幻想有一台完整的机器及其软件可供自己使用。这包括计算机支持的任何编程语言，以及用户想要使用的任何数据集，无论是由其他人提供还是由用户提供。唯一的限制是机器的物理极限。这远远超出了作为程序员工具的分时概念，也超出了 SAGE 的交互性质（它允许多个用户使用一个且只有一个数据集），也超出了 NASA 的实时系统（它将用户限制为只能使用专门的数据集和编程语言）。

The term “time-sharing” for computers has had several meanings. The earliest meaning was fairly restricted, and referred only to the ability of programmers to debug a program without having to prepare and submit a new deck of cards into the queue of jobs for a system. The time the computer spent going through the batch of jobs would be shared by a programmer making a few changes to a program, which would then be inserted into the original program that was already on the machine. By the late 1960s the term had a more general definition, one that was especially advocated by Professor John McCarthy, then of MIT, and which will be used in the following discussion of the System/360. By that definition, each user had the illusion that a complete machine and its software was at his or her disposal. That included whatever programming languages the computer supported, and any data sets the user wanted to use, whether supplied by others or by the user. The only constraint was the physical limits of the machine. That went far beyond the notion of time-sharing as a tool for programmers, as well as beyond the interactive nature of SAGE, which allowed multiple users of one and only one data set, and beyond NASA’s real-time systems, which restricted users to both specialized data sets and programming languages.

之所以能想到这一概念，是因为打字员的按键间隔至少几毫秒，而计算机能够获取并执行数十甚至数百条简单指令。打字员可能暂停几秒钟来思考下一个要输入的命令，而这几秒钟足以让计算机让另一个用户的作品出现、成长和消亡——至少在理论上是如此。实际上，指示计算机从一个用户切换到另一个用户的指令需要计算机处理器进行许多周期才能跟踪事物。这些指令所需的时间很容易占用用户按键之间的所有时间——甚至更多。但成功的分时系统带来的回报足以让许多人尝试构建一个。

What made such a concept thinkable was the disparity between the few milliseconds (at least) between a typist’s keystrokes and the ability of a computer to fetch and execute dozens, perhaps hundreds, of simple instructions. The few seconds a person might pause to ponder the next command to type in was time enough for a computer to let another user’s creation emerge, grow, and die—at least in theory. In practice, instructions that directed a computer to switch from one user to another required many cycles of the computer’s processor just to keep track of things. The time required by those instructions could easily take up all the time—and more—between a user’s keystrokes. But the rewards for a successful time-sharing system were great enough to lead many to try to build one.

到 20 世纪 60 年代中期，分时似乎是一个时机已到的想法。到 1962 年，麻省理工学院在 IBM 7090 上运行了一个实验系统。它使用升级后的 7094 演变成了兼容分时系统 (CTSS)。35 ^CTSS仅支持少数用户同时使用，但它确实成功解决了许多有关分时可行性的担忧。它在一定程度上促成了一项更雄心勃勃的系统提案，该系统将成为 MAC（“人与计算机”，又名“机器辅助认知”）项目的核心。36^在国防高级研究计划局的支持下，MAC 项目寻找一台计算机作为其系统的基础。System/360 是考虑的对象之一，但 1964 年 5 月，MAC 项目通知 IBM，刚刚发布的 System/360 并不合适。该项目选择了通用电气机器。³⁷此后不久，IBM 的另一个最受青睐的客户贝尔实验室抛弃了 IBM，选择 GE 系统进行分时工作。到 1967 年，GE 似乎已在计算机业务中占据领先地位，其成功的 GE-635 系列计算机似乎比 IBM 产品更能支持分时。

By the mid-1960s, time-sharing seemed an idea whose time had come. An experimental system was operating at MIT on an IBM 7090 by 1962. It evolved into Compatible Time-Sharing System (CTSS) using the upgraded 7094.³⁵ CTSS supported only a few users simultaneously, but it did successfully address many concerns about time-sharing’s viability. It led in part to a proposal for a more ambitious system, which would become the centerpiece for Project MAC (“Man and Computer,” a.k.a. “Machine-Aided Cognition”).³⁶ With support from the Defense Advanced Research Projects Agency, Project MAC sought a computer on which to base its system. System/360s were among those considered, but in May 1964 Project MAC informed IBM that the just-announced System/360 was not suitable. The project chose a General Electric machine instead.³⁷ Shortly after that, Bell Laboratories, another of IBM’s most favored customers, spurned IBM and chose a GE system for its time-sharing work. By 1967, GE seemed to be on its way to a position of leadership in the computer business, based on its successful GE-635 line of computers that seemed to support time-sharing better than IBM’s products.

IBM 自己对这些事件的描述描述了一种危机的气氛，一种感觉，经过如此艰苦的努力，公司正在迅速失去其在该领域的地位，并在竞争中沉沦。38 ³⁶⁰的基本架构并不反对分时应用程序，但也不是最佳的。39^一些MAC 项目参与者回忆说，System/360 最严重的缺陷是缺乏动态地址转换：能够停止程序的执行，将其从核心内存移到磁盘，然后在稍后将其移回（可能移到核心的不同部分）并恢复执行。MAC 项目团队的其他成员有不同的回忆——这与其说是动态地址转换的问题，不如说是 360 固有的“面向处理器”架构的问题。40^无论如何，如果每个用户都想让计算机的全部资源都用于解决他或她的问题，那么分时需要能够快速地将程序交换到核心和从核心交换程序。

IBM’s own history of those events describes an air of crisis, a feeling that after such hard work the company was rapidly losing its place in the field and sinking in the face of competition.³⁸ The 360’s basic architecture was not hostile to time-sharing applications, but neither was it optimum.³⁹ Some of the participants in Project MAC recall that the System/360’s most serious deficiency was its lack of dynamic address translation: an ability to stop the execution of a program, move it out of core memory to a disk, then at a later time move it back (probably into a different section of core) and resume execution of it. Other members of the Project MAC team have a different recollection—that it was not so much a matter of dynamic address translation as it was of the inherent “processor-oriented” architecture of the 360.⁴⁰ In any event, timesharing requires the ability to swap programs to and from core quickly, if each of several users is to have the illusion that the entire resources of the computer are brought to bear on his or her problem.

问题似乎在于，IBM 根本没有像一些客户希望的那样重视分时。阿姆达尔和布鲁克斯将注意力集中在引入统一处理器系列的必要性上，这可能导致他们忽视了 IBM 最佳客户对支持分时功能的内在硬件特性的深刻感受。阿姆达尔和布鲁克斯的回应是 360 可以满足这些客户的需求，这在技术上是正确的。但这并没有让麻省理工学院或贝尔实验室相信 IBM 与他们有着相同的计算未来愿景。

The problem seemed to be that IBM simply did not see time-sharing as as important as some customers wanted the company to see it. Amdahl and Brooks had focused their attention on the need to introduce a unified line of processors, which may have caused them to miss the depth of feeling among IBM’s best customers for intrinsic hardware features that supported time-sharing. Amdahl’s and Brooks’s response, that the 360 would meet these customers’ needs, was technically true. But it did not convince MIT or Bell Labs that IBM shared their vision of the future of computing.

IBM 以英勇地重新投入资源而闻名，该公司于 1965 年 8 月发布了带有地址转换硬件的 Model 67。历史学家认为 Model 67 是 IBM 的失败之作，也许是因为 IBM 的发布后来在反垄断审判中被作为证据。这一仓促的发布被认为是该公司所谓政策的一个例子，即发布产品更多的是为了切断竞争，而不是引入真正的价值。Model 67 的分时系统软件 (TSS) 运行不佳，IBM 的发布并没有阻止麻省理工学院和贝尔实验室继续从通用电气购买计算机。一位经历过那个时代的系统程序员说：

With the heroic rededication of resources that IBM was famous for, the company announced the Model 67 with address translation hardware in August 1965. Historians have accepted the view that the Model 67 was a failure for IBM, perhaps because IBM’s announcement of it was later introduced as evidence in an antitrust trial. This hasty announcement was cited as an example of the company’s alleged policy of announcing products more with an aim of cutting off competition than of introducing genuine value. The Model 67’s time-sharing system software (TSS) did not work well and IBM’s announcement of it did not stop MIT and Bell Labs from going through with their purchase of computers from General Electric. One systems programmer who lived through that era stated:

失去 MAC 项目和贝尔实验室对 IBM 产生了重大影响。此后，IBM 处理器很少成为尖端学术计算机科学研究的首选机器。MAC 项目将继续在 GE 645 上实现 Multics [分时操作系统]，并于 1969 年 10 月在麻省理工学院普遍使用。同样在 1969 年，贝尔实验室开始开发 UNIX 系统，作为 CTSS 和 Multics 的分支和优雅简化，该项目也不会使用 IBM 处理器^。41

Losing Project MAC and Bell Labs had important consequences for IBM. Seldom after that would IBM processors be the machines of choice for leading-edge academic computer science research. Project MAC would go on to implement Multics [a time-sharing operating system] on a GE 645 and would have it in general use at MIT by October, 1969. Also in 1969, the system that was to become UNIX would be begun at Bell Labs as an offshoot and elegant simplification of both CTSS and Multics, and that project, too, would not make use of IBM processors.⁴¹

这类评论既模糊又清晰。早期的分时系统运行良好，足以证明该概念的可行性，但需要数年时间才能实现其最初的承诺，即支持许多用户并以快速响应时间运行不同类型的作业。Model 67 虽然失败了，但为整个 System/360 系列的改造奠定了基础，其中分时可用并且运行良好：System/370 于 20 世纪末作为 360 系列的升级版发布。在整个 20 世纪 70 年代和 80 年代，System/370 使用对话监控系统 (CMS) 软件来处理要求苛刻的分时应用程序。

Comments like these are clouded as much as they are clarified by hindsight. Early time-sharing systems worked well enough to demonstrate the feasibility of the concept, but it took years before any of them fulfilled their initial promise of supporting many users and running different types of jobs with a quick response time. The Model 67, although a failure, laid the basis for a revamping of the entire System/360 line, in which time-sharing was available and did work well: the System/370, announced at the end of the decade as an upgrade to the 360 line. Throughout the 1970s and 1980s, System/370s using the Conversational Monitoring System (CMS) software handled demanding time-sharing applications.

通用电气也未能跃居领导地位。通用电气于 1970 年将其计算机业务出售给霍尼韦尔，这笔交易使霍尼韦尔在几年内获得了可观的利润并赢得了客户忠诚度。42^贝尔实验室发现 GE 分时系统存在不足，于 1969 年退出了 MULTICS 项目。那里的两位研究人员 Ken Thompson 和 Dennis Ritchie 最终开发了 UNIX，部分原因是在他们的雇主撤除 GE 系统后，他们需要一个工作环境。他们开始在数字设备公司的 PDP-7 上开发最终成为 UNIX 的东西，这台计算机的性能远不及 GE 大型机，在 1969 年就已过时。后来他们转移到 PDP-11。在接下来的十五年里，UNIX 的开发与 DEC 计算机联系在一起。这个名字意味着“UNIX”是“MULTICS”的简化形式，它借用了 MULTICS 的一些功能。 UNIX 也受到了早期、更简单的 CTSS 的启发。UNIX 对主流计算的影响发生在 20 世纪 80 年代，我们将在后面的章节中再次讨论。

Nor did General Electric vault to a position of leadership. GE sold its computer business to Honeywell in 1970, a sale that allowed Honeywell to make solid profits and gain customer loyalty for a few years.⁴² Bell Laboratories found the GE time-sharing system wanting, and dropped out of the MULTICS project in 1969. Two researchers there, Ken Thompson and Dennis Ritchie, ended up developing UNIX, in part because they needed an environment in which to do their work after their employer removed the GE system. They began work on what eventually became UNIX on a Digital Equipment Corporation PDP-7, a computer with far less capability than the GE mainframe and already obsolete in 1969. They later moved to a PDP-11. For the next decade and a half, UNIX’s development would be associated with DEC computers. The name implies that “UNIX” is a simplified form of “MULTICS,” and it did borrow some of MULTICS’s features. UNIX was also inspired by the earlier, and simpler, CTSS as well. UNIX’s impact on mainstream computing would occur in the 1980s, and it will be discussed again in a later chapter.

无论霍尼韦尔在获得通用电气的业务时拥有什么优势，这种优势都未能持续十年。我们将会看到，霍尼韦尔还浪费了小型计算机领域的领先地位。不过，霍尼韦尔和通用电气的损失并不是 IBM 的收获。成功的商业分时系统在 20 世纪 60 年代末变得很普遍——太普遍了，因为风险投资公司为市场提供了太多公司的资金。这些公司中的大多数既不使用 IBM 也不使用通用电气/霍尼韦尔的产品，而是使用数字设备公司的 PDP-10 或科学数据系统的 SDS-940。20 世纪 70 年代末和 80 年代初计算领域在交互性、网络和系统软件方面的革命性突破并非以 IBM 设备为中心。

Whatever advantage Honeywell had in obtaining GE’s business, it did not last the decade. We shall see that Honeywell also squandered a lead in minicomputers. Still, Honeywell’s and GE’s loss was not IBM’s gain. Successful commercial time-sharing systems became common in the late 1960s—too common, as venture capital firms funded too many companies for the market. Most of these companies used neither IBM nor GE/ Honeywell products, but rather PDP-10s from Digital Equipment Corporation or SDS-940s from Scientific Data Systems. The revolutionary breakthroughs in interactivity, networking, and system software that characterized computing in the late 1970s and early 1980s would not be centered on IBM equipment.

System/360 Model 67 重复了 IBM 的 STRETCH 经历。两者都是商业上的失败，但都为随后的成功产品线奠定了基础，这些产品的成功弥补了最初的收入损失。诉讼中提出的问题掩盖了 System/360 架构的教训——因为它设计得非常好，所以它可以在五年后吸收重大改进，同时仍然保持软件兼容性和客户群。至于麻省理工学院投奔通用电气所引发的危机，可以说 IBM 应该对麻省理工学院更加包容。公平地说，事后看来，麻省理工学院最好将 MAC 项目建立在与其选择的计算机不同的计算机上。

The System/360 Model 67 repeated IBM’s experience with STRETCH. Both were commercial failures, but both laid the groundwork for the line of successful products that followed, whose success overcame the initial lost revenues. The issues raised in litigation have obscured the lesson of System/360 architecture—because it was so well designed, it could absorb a major enhancement five years later and still maintain software compatibility and its customer base. As for the crisis precipitated by MIT’s defection to General Electric, one can say that IBM should have been more accommodating to MIT. To be fair, one could also say, in hindsight, that MIT would have been better off basing Project MAC around a different computer from the one it chose.

随着 20 世纪 60 年代的结束，小型计算机（尤其是 PDP-11 和 Data General Nova 等新型号）开始走出实验室和 OEM 环境，进入 IBM 和“七个小矮人”认为属于自己的领域。不幸的是，对于 IBM 来说，360 的架构不允许推出小型计算机版本 — 低端 360 Model 20 是 IBM 可以推出的最小最便宜的产品，而且它已经与该系列的其他产品部分不兼容。一些基本的架构特性，尤其是输入/输出通道的使用以及磁带和磁盘驱动器的标准化接口，阻碍了价格大幅下降。

As the 1960s drew to a close, the minicomputer—especially new models like the PDP-11 and Data General Nova—was breaking out of its laboratory and OEM setting and moving into territory that IBM and the “Seven Dwarfs” had considered their own. Unfortunately for IBM, the 360’s architecture did not permit a minicomputer version—the low-end 360 Model 20 was about as small and cheap as IBM could go with the line, and it was already partly incompatible with the rest of the line. Some basic architectural features, especially the use of input/output channels and standardized interfaces to tape and disk drives, prevented going much lower in price.

因此，尽管该系列的名称如此，但 System/360 承诺的整个产品线中仍缺少一些领域 — — 这些领域将在未来二十年内增长。360 的订单使 IBM 的资源紧张，尽管 IBM 仍然足够大，足够精明，不会忽视小型计算机市场，并于 1969 年推出了 System/3，这是一款不兼容的计算机，每月租金低至 1000 美元。43 ^“坦率地承认 System/360 无法弥补市场上不断扩大的机会。” ⁴⁴ System/3 是一款成功的产品，因为它比小型计算机更容易适应小型企业和会计工作，而小型计算机通常带有实验室工作台的传统。System/3 的一个有趣特性是它使用了一种新的、更小的穿孔卡，可以编码 96 个字符。即使在推出这款产品时，IBM 也在开发将永远取代穿孔卡的存储介质：软盘。廉价磁盘存储设备一经面世，客户便抛弃了不兼容的打孔卡。然而，古老的 80 列打孔卡在接下来的十年里依然很受欢迎。

So despite the name of the series, there were sectors missing from the full circle promised by System/360—sectors that would grow in the next two decades. Orders for the 360 strained IBM’s resources, although IBM was still big enough and astute enough not to ignore the minicomputer market, and in 1969 it responded with System/3, an incompatible computer that could be rented for as low as $1000 a month.⁴³ It was “a candid concession that System/360 could not bridge the widening opportunities in the marketplace.”⁴⁴ System/3 was a successful product, because it was easier to adapt for small business and accounting jobs than minicomputers, which often carried their laboratory workbench heritage with them. One interesting feature of the System/3 was its use of a new and smaller punched card that could encode 96 characters. Even as this product was introduced, IBM was developing the storage medium that would forever displace the punched card: the floppy disk. As soon as inexpensive disk storage was made available for the computer, customers abandoned the incompatible punched card. The venerable 80-column card, however, continued to be popular through the next decade.

《财富》杂志曾说“你把公司押注在” 360 上，这让 IBM 的高层感到不安，事实证明这是真的，但原因却错了。威胁公司生存的是成功，而不是失败。成功意味着需要迅速筹集资金、建造新工厂、雇用新员工并扩大生产。这意味着任何希望公司采用另一种计算方式（如 MAC 项目或小型计算机公司推广的计算方式）的人都必须逆 360 订单的潮流而行。

Fortune magazine’s quote that “you bet your company” on the 360, which disturbed IBM’s senior management, turned out to be true, but for the wrong reason. It was success, not failure, that threatened the company’s existence. Success meant the need to raise capital quickly, build new plants, hire new workers, and expand production. It meant that anyone who wanted the company to adopt an alternate style of computing, such as the ones that Project MAC or the minicomputer companies promoted, would have to swim against the rushing current of 360 orders.

股市飙升时期

The Period of Soaring Stocks

受国防部为越南战争投入的资金以及 NASA 为了将人类送上月球而对计算能力的无限渴求的刺激，20 世纪 60 年代末是美国计算机行业蓬勃发展的时期。对于那些还记得 20 世纪 80 年代个人电脑爆炸式增长的人来说，很容易忽略更早的事件。从 1966 年到 1968 年，几乎所有名称中带有“-ex”、“-tronics”或“-tron”的股票都呈上升趋势，可与登月火箭相媲美。45^著名商业记者、敏锐的华尔街观察家约翰布鲁克斯 (John Brooks) 将其称为“繁荣时期”，指的是经纪人们随着每日股票行情的上涨而狂热地高呼^。46

Spurred on by Defense Department spending for the Vietnam War, and by NASA’s insatiable appetite for computing power to get a man on the Moon, the late 1960s was a time of growth and prosperity for the computer industry in the United States. For those who remember the personal computer explosion of the 1980s, it is easy to overlook earlier events. From about 1966 to 1968, almost any stock that had “-ex,” “-tronics,” or simply “-tron” in its name rode an upward trajectory that rivalled the Moon rockets.⁴⁵ John Brooks, a well-known business journalist and astute observer of Wall Street, labeled them the “go-go years,” referring to the rabid chants of brokers watching their fortunes ascend with the daily stock ticker.⁴⁶

一些高涨的股票是由专注于晶体管和新发明的集成电路等基础技术的公司发行的。其他一些则是追随数字设备公司脚步的全新小型计算机公司。其他一些则是分时公用事业公司。其他一些则是软件和服务公司，它们的出现是为了帮助客户轻松应对复杂而昂贵的新型大型机的操作。一些是“插件兼容制造商”，它们以较低的成本销售与 IBM 产品线兼容的系统部件。最后，还有一些第三方租赁公司，它们生活在 IBM 定价和租赁政策的“保护伞”之下。我们将从后者开始。

Some of the go-go stocks were issued by companies that focused on the base technology of transistors and the newly invented integrated circuit. Others were brand-new minicomputer companies following in Digital Equipment Corporation’s footsteps. Others were the time-sharing utilities. Others were software and service companies, which sprouted to help ease customers into the complexity of operating the complex and expensive new mainframes. Some were “plug-compatible manufacturers,” which sold, at lower cost, pieces of a system that were compatible with IBM’s product line. Finally, there were third-party leasing companies that lived under an “umbrella” of IBM’s pricing and leasing policy. We shall begin with the latter group.

租赁公司1956 年，IBM 与美国政府达成协议，同意出售和租赁其计算机。然而，租赁仍然占据主导地位，因为 IBM 拥有其他公司所没有的资本。许多客户喜欢租赁，因为租赁不会占用他们的资金，而且如果出现新技术，他们可以取消租赁并转向更好的产品。IBM 则从租赁中获得了稳定的现金流，尽管它必须应对竞争对手提供采用新技术的机器的挑战。

Leasing Companies Having entered into a consent decree with the U.S. government in 1956, IBM agreed to sell as well as lease its computers. Leasing continued to predominate, however, because IBM had the capital that few others had. And many customers liked leasing, which did not tie up their capital, and made it possible to cancel a lease and move to the better product if new technology came along. IBM, in turn, received a steady flow of cash from its leases, although it had to meet the challenge of competitors offering machines with newer technology.

20 世纪 60 年代中期租赁公司兴起的关键在于人们认为 IBM 对其设备收取了过高的租金，具体来说，IBM 收取租金是基于这样的预期：计算机将在短短五年内过时，因此需要偿还。租赁公司认为客户会希望将计算机保留更长时间，最长可达十年。从 1961 年在布鲁克林阁楼成立的 Leasco 公司开始，这些公司从 IBM 购买大型机（根据同意令），然后以比 IBM 收费低 20% 的价格出租给客户。47这些公司很^容易就从华尔街筹集到了必要的资金，因为他们能够让金融家相信，IBM 收取的高价为他们提供了保护。

The key to the emergence of leasing companies in the mid-1960s was the perception that IBM was charging artificially high rents on its equipment—specifically, that IBM was charging rent on the expectation that the computers would become obsolescent, and therefore returned from lease, in as little as five years. The leasing companies reckoned that customers would want to hang on to them longer—up to ten years. Beginning with a company called Leasco that started in a Brooklyn loft in 1961, these companies would buy mainframes from IBM (as permitted by the consent decree), and then rent them to customers at up to 20 percent less than what IBM charged.⁴⁷ These companies had little trouble raising the necessary capital from Wall Street, as they were able to convince financiers that they were sheltered under an umbrella of artificially high prices charged by IBM.

问题仍然是 IBM 的价格是否真的太高。鉴于推出 System/360 的成本巨大，IBM 自然希望收回这笔投资。然而，还有其他因素。System/360 为客户提供了迁移到新计算机的自己的软件兼容途径。在这一代大型机中，编写软件的成本比以往任何时候都高，这可能会鼓励客户继续使用计算机，即使竞争对手提供更好的硬件但没有软件兼容性。预测客户将保留大型机多长时间的财务模型必须考虑到这一点，但要考虑到多少呢？

The question remained whether IBM’s prices were, in fact, too high. Given the enormous cost of bringing out System/360, IBM naturally wanted to get that investment back. There were other factors, however. System/360 offered its own software-compatible path for a customer to migrate to a new computer. With this generation of mainframes, the cost of writing software was higher than ever before, which might encourage customers to hold onto a computer even if a competitor offered better hardware but no software compatibility. Financial models that predicted how long a customer might keep a mainframe had to take this into account, but by how much?

就在 360 即将交付给第一批客户时，IBM 就意识到技术进步的步伐不会减缓。该公司在研究设施上投入了大量资金，在纽约约克敦高地、加利福尼亚圣何塞和欧洲建立了实验室。这是租赁公司不必承担的费用，但这意味着 IBM 可以确保至少一些技术创新在其控制之下。在 IBM 主导计算机行业的情况下，它可以逐步推出这种创新，从而不会使其安装基础过快过时。但主导地位转瞬即逝：如果 IBM 过于保守，另一家公司肯定会进入。其他公司确实这样做了，正如 Control Data 和 RCA 的例子所揭示的那样。

Even as the 360 was going to its first customers, IBM recognized that the pace of technology was not going to slacken. The company spent lavishly on research facilities, building up laboratories in Yorktown Heights, New York, San Jose, California, and in Europe. That was an expense the leasing companies did not have to bear, but it meant that IBM could ensure that at least some technical innovation was under its control. To the extent that IBM dominated the computer industry, it could mete out this innovation gradually, thus not making its installed base obsolete too quickly. But dominance was a fleeting thing: if IBM held back too much, another company was sure to enter in. And others did, as the examples of Control Data and RCA will reveal.

IBM 究竟控制了多少市场成为了 1969 年开始的另一项联邦反垄断诉讼的主题，但没有一家公司，无论规模有多大，能够控制底层技术的发展速度。通用电气向需要分时设施的客户销售产品只是一个例子。无论如何，在 20 世纪 60 年代末，IBM 宣布了 System/360 的继任者 — — 370 系列。该系列与 360 在软件上兼容，但更适合分时，并且使用集成电路作为处理器和内存。System/370 的推出可能比 IBM 预期的要早，但到了 1970 年，它别无选择。48^考虑到 System/370 发布的时间，也许 IBM 的价格毕竟不算太高。

Just how much of the market IBM controlled became the subject of another federal antitrust action beginning in 1969, but no single company, no matter how big, could control the pace of the underlying technology. GE’s sales to customers who wanted time-sharing facilities was only one example. In any event, at the end of the 1960s IBM announced a successor to System/360—the 370 line. This line was software-compatible with the 360, but it was better suited for timesharing, and it used integrated circuits for its processor and memory. System/370 probably came sooner than IBM wanted, but by 1970 it had no choice.⁴⁸ Given the timing of System/370’s announcement, perhaps IBM’s prices were not too high after all.

20 世纪 70 年代末，该公司还不得不应对其他压力。这些压力来自小型计算机公司，这些公司的产品正在发展以处理大型机应用程序。1978 年，IBM 推出了一款低成本的 4300 系列，与 360-370 系列兼容，以及一款不兼容的中型 AS/400。为了应对大型机制造商的压力，它推出了 370 大型机型的继任者 — — 3030 系列。这些竞争压力 — — 来自其他大型机、小型计算机、分时系统和低成本工作站（稍后讨论） — — 都不足以造成严重损害，但它们的结合 — — 到 1990 年发展成为廉价工作站的网络系统 — — 足以造成严重损害。

In the late 1970s the company had to respond to other pressures as well. These came from the minicomputer companies whose products were evolving to handle mainframe applications. In 1978, IBM introduced a low-cost 4300-series, compatible with the 360-370 line, and a midsized AS/400, which was not compatible. It responded to pressures from the makers of large mainframes by bringing out a successor to the larger models of the 370—the 3030 series. None of these competitive pressures—from other mainframes, from the minicomputer, from timesharing, and from low-cost workstations (discussed later)—were enough to do serious damage, but their combination, evolving by 1990 into a networked system of inexpensive workstations, would.

System/360 在市场上的主导地位肯定比租赁公司押注的十年要短，尽管并非每个客户都觉得需要立即升级。无论如何，技术的发展速度，加上 1971 年牛市的结束，以及 IBM 对定价和产品公告的精心操纵，导致定价保护伞在 20 世纪 70 年代中期关闭，租赁公司的投资者蒙受了巨大的财务损失。

System/360’s dominance of the market was certainly shorter than the ten years that the leasing companies had gambled on, although not every customer felt the need to upgrade immediately. In any event, the pace of technology, combined with the end of the bull market in 1971 and with IBM’s careful manipulation of pricing and product announcements, served to fold up the pricing umbrella by the mid-1970s, leaving investors in the leasing companies with heavy financial losses.

兼容大型机System/360 发布后的第二个后果是重新定义了 IBM 主要竞争对手的角色，并导致一些小型公司直接瞄准 360 系列。在 SPREAD 委员会的最初讨论中，有人表示担心，通过推出广泛的机器系列，“竞争对手会拿着步枪追赶每一种 [特定型号]。”不受兼容性需求的限制，有人可以以相同的成本推出性能更好的机器。49^委员会不得不争辩说，为 360 客户提供向上迁移的途径的优势将压倒竞争对手瞄准特定型号的任何优势。

Compatible Mainframes The second consequence of the announcement of System/360 was a redefinition of the role of IBM’s principal competitors, and led to the emergence of smaller companies aimed directly at the 360 line. During the initial SPREAD Committee discussions, someone expressed the fear that by introducing a broad line of machines, “the competition would be out after each [specific model] with a rifle.” Unconstrained by a need for compatibility, someone could bring out a machine with far better performance for the same cost.⁴⁹ The committee had to argue that the advantages of having a path for upward migration for 360-customers would overwhelm any advantages of a competitor’s shot at a particular model.

无论如何，在最高端，没有更高的型号可以迁移到。推出了前面提到的小型 160A 的控制数据公司 (Control Data Corporation) 于 1964 年推出了其 6600 计算机（图 5.4）。6600 由西摩·克雷 (Seymour Cray) 设计，很快被称为“超级计算机”，它提供了西摩·克雷 (Seymour Cray) 想要在计算机中提供的功能 — 最快的性能。50^就绝对销量而言，CDC 6600 并不构成太大威胁，但它的客户很不寻常：劳伦斯利弗莫尔 (Lawrence Livermore) 等武器实验室、大型空气动力学研究组织、国家安全局 (National Security Agency) 以及其他只关心性能的机构。这些客户可能总共只购买几台，但其他不那么光鲜的客户却对他们评价很高。因此，无论他们选择什么系统，行业媒体都会认真报道和讨论。 IBM 推出了 System/360（型号 91），但交付较晚，而且其性能始终无法与 CDC 机器相媲美。

At the highest end there would be no higher model to migrate to anyway. Control Data Corporation, which introduced the small-scale 160A already mentioned, came out with its 6600 computer in 1964 (figure 5.4). Designed by Seymour Cray and soon dubbed a “supercomputer,” the 6600 offered what Seymour Cray wanted to deliver in a computer—the fastest performance possible, period.⁵⁰ In terms of absolute sales, the CDC 6600 was not much of a threat, but its customers were unusual: the weapons laboratories like Lawrence Livermore, the large aerodynamics research organizations, the National Security Agency, and others for whom performance was all that mattered. These customers might collectively buy only a few units, but other, less glamorous customers held them in high regard. Whatever systems they chose was therefore reported and discussed seriously in the trade press. IBM countered with a System/360, Model 91, but it was late in delivering it, and its performance never matched the CDC machines.

图 5.4

控制数据公司 CDC-6600 的控制台，约 1964 年。6600 由 Seymour Cray 设计，并使“超级计算机”一词流行起来。（来源：控制数据公司档案。）

Figure 5.4

Console of the Control Data Corporation’s CDC-6600, ca. 1964. The 6600 was designed by Seymour Cray and popularized the term “supercomputer.” (Source : Control Data Corporation Archives.)

兼容性问题在这里发挥了作用。6600 的性能很大程度上归功于 Seymour Cray 的设计天赋，而 Cray 的另一个优势是，无论何时设计计算机，他都可以“从一张白纸开始” — 不受兼容性问题的限制。51 ^IBM在高端市场之争中败给了 CDC（后来又败给了 Seymour Cray 于 1972 年创立的 Cray Research）。Control Data Corporation 最终起诉 IBM，称 Model 91 是“幻影”，在准备就绪之前就宣布了，目的是扼杀 CDC 6600。无论诉讼的优劣如何，除了少数例外客户外，大多数人都更喜欢软件兼容性的优势。甚至许多拥有无限预算的武器实验室也在 CDC 6600 旁边安装了一台或多台 System/360。

Here the compatibility issue took its toll. The 6600 owed a lot of its performance to Seymour Cray’s talent for design, and Cray had the further advantage that he could “start with a clean sheet of paper”— unconstrained by compatibility issues—whenever he designed a computer.⁵¹ IBM lost the battle for the high end to CDC (and later on to Cray Research, founded by Seymour Cray in 1972). Control Data Corporation eventually sued IBM, alleging that the Model 91 was a “phantom,” announced before it was ready, in order to kill the CDC 6600. Whatever the merits of the lawsuit, except for a few exceptional customers, most preferred the advantages of software compatibility. Even many weapons laboratories, with their unlimited budgets, installed one or more System/360s alongside their CDC 6600.

360 的另一个威胁来自 RCA。RCA 曾在 20 世纪 50 年代尝试过 BIZMAC 进入商业计算领域，但未能成功，但在 1959 年，RCA 凭借 501 型取得了更好的成绩，这是一款小型商用计算机，以率先配备 COBOL 编程语言编译器而闻名。普渡大学的 Saul Rosen 教授曾说过：“它非常慢，而且……COBOL 编译器也非常慢，但对于许多用户来说，慢速 COBOL 比没有 COBOL 要好。” ⁵²另一台晶体管计算机 301 型能够实时操作。（NASA 使用一台 301 型来对将人类送上月球的土星火箭进行排序和控制。）在 20 世纪 50 年代末和 60 年代初，该公司专注于将彩色电视带入消费市场，而数字电子产品的优先级较低。

Another threat to the 360 came from RCA. RCA had tried and failed to enter commercial computing in the 1950s with the BIZMAC, but in 1959 RCA had better luck with the model 501, a small commercial computer that was best known for being one of the first to be supplied with a compiler for the COBOL programming language. Professor Saul Rosen of Purdue University once said, “It was quite slow, and ... the COBOL compiler was also very slow, but for many users a slow COBOL was better than no COBOL.”⁵² Another transistorized computer, the Model 301, was capable of real-time operation. (One was used by NASA to sequence and control the Saturn rockets that took men to the Moon.) Through the late 1950s and early 1960s, the company focused on bringing color television to a consumer market, giving digital electronics a lower priority.

1964 年末，RCA 宣布了一项更大胆的攻势，即 Spectra 70 系列。这是由四台计算机组成的产品线，它们无需修改即可执行为 IBM 360 系列相应型号编写的软件。而且它们的价格最多可降低 40%。这正是 IBM 所担心的竞争。RCA 无需在 IBM 实验室内安插间谍——他们可以依靠提供给客户和软件开发人员的规格。System/360 项目规模如此之大，以至于 IBM 不得不分享大量相关信息，而且它无力阻止某人制造后来被称为“克隆”的产品。制造兼容 360 的计算机也成为苏联快速构建强大大型机的途径。

Late in 1964 RCA announced a bolder offensive, the Spectra 70 series. This was a line of four computers that would execute, without modification, software written for corresponding models of the IBM 360 line. And they would cost up to 40 percent less. This was the competition IBM had feared. RCA did not need to plant spies inside IBM’s laboratories— they could rely on specifications supplied to customers and software developers. The System/360 project was so big that IBM had to share a lot of information about it, and it was powerless to stop someone from building what later on would be called a “clone.” Building 360-compatible computers also became a quick way for the Soviet Union to construct powerful mainframes.

RCA 的激进定价源自多个因素。由于它省去了 IBM 承担的架构设计成本，其开发成本不到 IBM 的十分之一。53^而且由于起步较晚，RCA 还可以利用组件技术的进步。两款机型 Spectra 70/45 和 70/55 使用了真正的集成电路，因此性价比比 360 更高。

RCA’s aggressive pricing came from several factors. Because it had escaped the cost of designing the architecture that IBM had borne, its development costs were less than one tenth of IBM’s.⁵³ And by starting later, RCA could also take advantage of advances in component technology. Two models, the Spectra 70/45 and the 70/55, used true integrated circuits, and thus offered better performance for the dollar than the 360.

RCA 的 Spectra 70 系列大获成功，但在 IBM 于 1970 年推出同样使用集成电路的 System/370 进行反击后，销量开始下滑。在遭受巨大损失且看不到尽头后，RCA 于 1971 年宣布退出计算机业务。以低价，安装的客户群被 UNIVAC（现为 Sperry 的一个部门）收购，而数字设备公司则收购了 RCA 位于马萨诸塞州马尔伯勒的全新制造厂。UNIVAC 继续为 RCA 机器提供服务，并精心培育拥有这些机器的数百家公司，最终让它们转向 UNIVAC 大型机。这与 1970 年通用电气将计算机业务出售给霍尼韦尔的情况如出一辙：一家电子巨头将业务出售给一家更了解商业设备营销的公司。Sperry UNIVAC 获得了一笔好交易。它只为 RCA 的业务支付了 7000 万美元，而霍尼韦尔为通用电气支付了 2.34 亿美元。⁵⁴数字设备公司在万宝路工厂获得了“一生难求的交易”，发现那里的现代化设施正是其快速扩张时期所需要的^。55

RCA’s Spectra 70 Series was successful, but sales withered after IBM returned fire in 1970 with its System/370, which also used integrated circuits. After incurring massive losses with no end in sight, RCA announced in 1971 that it was leaving the computer business. For a bargain price, the installed customer base was bought by UNIVAC (now a division of Sperry), and Digital Equipment Corporation bought RCA’s brand-new manufacturing plant in Marlboro, Massachusetts. UNIVAC continued to service the RCA machines and carefully cultivated the hundreds of companies that owned them, eventually easing them over to UNIVAC mainframes. This echoed the sale of GE’s computer business to Honeywell in 1970: an electronics giant selling out to a company more knowledgeable about marketing business equipment. Sperry UNIVAC got a good deal. It paid only $70 million for RCA’s business, compared to the $234 million that Honeywell had paid for GE’s.⁵⁴ Digital Equipment Corporation got the “deal of a lifetime” in the Marlboro plant, finding its modern facilities just what it needed at a time when it was expanding rapidly.⁵⁵

插头兼容制造商

The Plug-Compatible Manufacturers

RCA 的失败并没有否定复制 360 架构的基本经济效益。其他资本远少于 RCA 的公司也取得了成功，他们不是通过复制整个产品线，而是通过瞄准 360 系统的各个部分：内存单元、磁带驱动器和中央处理器。这些公司也在 IBM 定价政策的保护下运营，建立了“即插即用制造商”或 PCM 业务，这是繁荣时期的另一个决定性部分。

RCA’s failure did not invalidate the basic economics of copying the 360 architecture. Other companies with far less capital than RCA proved successful, not by copying the entire line, but by targeting pieces of the 360 system: memory units, tape drives, and central processing units. These companies, also operating under the umbrella of IBM’s pricing policies, established the “plug-compatible manufacturer,” or PCM business, another defining segment of the go-go years.

1970 年，IBM 的明星计算机设计师之一 Gene Amdahl 离开 IBM，创办了一家生产兼容处理器的公司。Amdahl 公司于 1975 年开始安装其 Model 470 V/6，这款机器与 CDC-6600 一样，可以与 IBM 的顶级产品相媲美。不仅如此，它的性能远远超过后者。与 CDC 6600 不同，Amdahl 处理器可以运行 IBM 360 软件。56^在此之前，Memorex、Telex、Ampex、Storage Technology 和 CalComp 等公司都在提供磁带驱动器、磁盘驱动器甚至主内存单元，人们只需将它们插入 IBM 360 或 370 安装即可取代 IBM 自己的设备（因此得名），以更低的价格提供相同或更好的性能。与 Amdahl 处理器相结合，人们可以构建一个完整的计算机系统，几乎无需任何“蓝色巨人”硬件即可运行所有 360 软件。（IBM 的 360 产品被漆成了独特的蓝色。）

In 1970, Gene Amdahl, one of the company’s star computer designers, left IBM to found a company that would make a compatible processor. The Amdahl Corporation began installations in 1975 of its Model 470 V/6, a machine that, like the CDC-6600, competed with the top of the IBM line. More than that, it far outperformed it. Unlike the CDC 6600, the Amdahl processor could run IBM 360 software.⁵⁶ Well before that time, companies like Memorex, Telex, Ampex, Storage Technology, and CalComp were offering tape drives, disk drives, and even main memory units, that one could simply plug into an IBM 360 or 370 installation in place of IBM’s own equipment (hence the name), giving equal or better performance at a lower price. Coupled with an Amdahl processor, one could thus build a complete computer system that ran all the 360 software almost without any “Big Blue” hardware. (IBM’s 360 products were painted a distinctive blue.)

这些才是弗雷德·布鲁克斯所担心的真正的“枪击”，它们造成了巨大的损失。在华尔街，这些公司的股票飙升。在某些情况下，IBM 的回应是重新包装和重新定价其产品，以缩小成本差异。这些行动在 1969 年至 1975 年间引发了不少于10 起诉讼，指控 IBM 违反反垄断法。57^除此之外，美国司法部于 1969 年发起了反垄断诉讼，以及前面提到的 Control Data 就 Model 91 提起的诉讼。大多数诉讼在 1980 年之前得到解决，那时技术的发展已经使诉讼内容变得无关紧要。

These were the true “rifle shots” that Fred Brooks had worried about, and they did a lot of damage. On Wall Street, stock in these companies soared. IBM responded in some cases by repackaging and repricing its products to make the cost difference less. Those actions spawned no fewer than ten lawsuits between 1969 and 1975, charging IBM with antitrust violations.⁵⁷ This was in addition to the U.S. Justice Department’s own antitrust suit, launched in 1969, and the Control Data lawsuit over the Model 91 already mentioned. Most of these suits were settled by 1980, by which time the rush of technology had rendered their substance irrelevant.

尽管 20 世纪 70 年代的熊市对其造成了冲击，但一些插拔兼容机仍然蓬勃发展。阿姆达尔幸存下来，主要得益于富士通的融资，富士通是一家日本公司，一直在寻找进入美国计算机市场的途径。1975 年，富士通开始在日本制造阿姆达尔计算机并在美国销售，依靠吉恩·阿姆达尔的才能提供基于最新集成电路技术的有竞争力的设计。另一家日本公司日立也开始制造和销售插拔兼容主机，并以美国国家半导体公司的子公司 National Advanced Systems 的名义在美国销售。58^这是美国计算机行业面临的第一个来自日本的严峻竞争。作为 PCM 现象的副产品，它比大多数 PCM 公司本身存在的时间更长。 IBM 的反击是 20 世纪 70 年代末推出的更快的 3030 系列，这减缓了客户转而使用 Amdahl 机器的速度，留下了一个日本稳固占据主导地位的竞争环境。1980 年，Gene Amdahl 离开了仍以他的名字命名的公司，成立了 Trilogy，但 Trilogy 从未取得他所希望的成功。许多磁带和磁盘制造商倒闭或合并；很少有制造商能存活到工作站时代^。59

Some of the plug-compatibles prospered throughout all this, although the bear market of the 1970s took its toll. Amdahl survived, mainly through financing from Fujitsu, a Japanese company that had been casting about for ways to enter the U.S. computer market. In 1975 Fujitsu began building Amdahl computers in Japan for sale in the United States, relying on Gene Amdahl’s talent to offer a competitive design based on the latest integrated-circuit technology. Another Japanese company, Hitachi, began making and selling plug-compatible mainframes as well, sold in the United States under the name National Advanced Systems, a division of National Semiconductor Corporation.⁵⁸ This was the first serious Japanese competition ever faced by the U.S. computer industry. A by-product of the PCM phenomenon, it long outlasted most of the PCM companies themselves. IBM’s counter punch, the faster 3030-series introduced in the late 1970s, slowed the defections of customers to Amdahl’s machines, leaving a competitive environment with the Japanese firmly entrenched. Gene Amdahl left the company still bearing his name to form Trilogy in 1980, but Trilogy never achieved the success he had hoped for. Many of the tape and disk manufacturers folded or merged with one another; few survived into the workstation era.⁵⁹

UNIVAC，SDS System/360 的另一个后果是，几家竞争对手在整个圈子中发现 360 覆盖范围不广的小部分。除了覆盖高端和低端的超级计算机和小型计算机外，在中间，一些科学和工程客户发现，与他们不得不放弃的优雅的 7090 相比，中端 360 的性能受到影响。这些客户比其他客户更愿意选择竞争对手，主要是因为他们习惯于编写自己的大部分软件。Sperry UNIVAC 开发了其 1100 系列大型机的一个版本，该版本采用集成电路并提供快速处理速度。1964 年发布的 1108 大型机销量特别好。60¹¹⁰⁰系列计算机对 360 系列构成了强烈挑战，是 UNIVAC 大部分利润的基础，直到 1970 年代初。对联邦航空管理局等政府和军事机构的销售尤其强劲^。61

UNIVAC, SDS Another consequence of the System/360 was that several competitors found small sectors in the full circle where the 360’s coverage was spotty. Besides the supercomputers and minicomputers covering the high and low ends, in the middle some scientific and engineering customers found that the performance of mid-range 360s suffered in comparison to the elegant 7090s they were obliged to abandon. These customers were more willing than others to go with a competitor, primarily because they were used to writing much of their own software. Sperry UNIVAC developed a version of its 1100 series of mainframes that employed integrated circuits and offered fast processing speeds. The 1108 mainframe, announced in 1964, sold especially well.⁶⁰ The 1100-series of computers was a strong challenge to the 360 line and were the basis for most of UNIVAC’s profits into the early 1970s. Sales were especially strong to government and military agencies such as the Federal Aviation Administration.⁶¹

更戏剧性的是与 CDC 和 DEC 同时成立的一家公司的成功，即加州科学数据系统公司。SDS 的创始人是马克斯·帕列夫斯基，他是一名哲学专业的大学学生，他发现，由于数字计算机的发明，20 世纪 50 年代是“历史上哲学家第一次找到工作！” ⁶²帕列夫斯基最初在本迪克斯从事计算机项目，然后加入了一家小型电子公司 Packard-Bell，他声称自己是在黄页上发现这家公司的。他说服该公司应该进入计算机行业。他作为哲学家的训练不仅帮助他理解了计算机逻辑，还帮助他理解了辩论的艺术。结果，廉价的 PB-250 在 1960 年推出时取得了一定的成功。63^{帕列夫斯基}与二战后来到阿拉巴马州亨茨维尔的德国火箭工程师建立了特别良好的关系，他们在那里参与了陆军的弹道导弹计划。帕列夫斯基在本迪克斯研究的计算机虽然设计新颖，但却引起了这个正在寻找实时控制导弹方法的团队的注意。

More dramatic was the success of a company that started up at the same time as CDC and DEC, namely Scientific Data Systems of California. SDS was founded by Max Palevsky, a philosophy major in college who found that, because of the invention of the digital computer, the 1950s was the “first time in history that a philosopher could get a job!”⁶² Palevsky had first worked on a computer project at Bendix, then joined Packard-Bell, a small electronics firm that he claims he found in the Yellow Pages. He convinced the company that they ought to enter the computer business. His training as a philosopher helped him understand not only computer logic but also the art of argumentation. The result, the inexpensive PB-250, was a modest success when it was introduced in 1960.⁶³ Palevsky developed an especially good relationship with the German rocket engineers who had come after World War II to Huntsville, Alabama, where they were involved with the Army’s ballistic missile program. The computer that Palevsky had worked on at Bendix, though of an unconventional design, had caught the attention of this group, who were looking for ways of controlling a missile in real time.

1961 年，帕列夫斯基离开 Packard Bell，创立了自己的公司 Scientific Data Systems，筹集了约 100 万美元的风险投资。不到一年，SDS 就推出了一款计算机 Model 910，公司实现了盈利。帕列夫斯基将 910 的成功归功于其卓越的输入/输出功能。它还有效利用了组件技术的最新发展成果。910 是首批使用硅晶体管的非军用计算机之一，1965 年交付的型号是首批（与 RCA 一起）使用集成电路的计算机之一^。64

In 1961 Palevsky left Packard Bell and founded his own company, Scientific Data Systems, raising around $1 million in venture capital. Within a year SDS had introduced a computer, the Model 910, and the company was profitable. Palevsky attributes the 910’s success to its superior Input/Output facilities. It also made effective use of the latest developments in component technology. The 910 was one of the first nonmilitary computers to use silicon transistors, and a model delivered in 1965 was one of the first (along with RCA’s) to use integrated circuits.⁶⁴

到 1964 年，SDS 的收入超过了 DEC——当然，它开始时的资本是 DEC 的十倍多。1969 年，当 SDS 的销售额达到 1 亿美元，税后收入达到 1000 万美元时，Palevsky 以价值 9 亿美元的施乐股票将公司卖给了施乐。65 ^（几年后，施乐注销了该部门，损失了 13 亿美元。）由于起步不大，SDS 经常被拿来与 DEC 比较，它的计算机有时被归类为小型计算机。12 位 Model 92 在某些方面类似于 PDP-8，但 SDS 的主要业务是 24 位 910 和 920——大型科学计算机，不如 Control Data 的超级计算机强大，但比小型计算机功能强大得多。 1965 年推出的另一款机型⁹⁴⁰明确针对分时使用进行营销，并且同样取得了成功。66许多西海岸分时公司（包括先驱 Tymshare）都以它为基础。940 被用于加利福尼亚州帕洛阿尔托地区（在被称为“硅谷”之前）许多人机交互和网络方面的开创性研究项目。从某种意义上说，它是 DEC PDP-10 的西海岸对应物，DEC PDP-10 是一款精心设计的分时系统，对高级计算研究产生了最大影响。通过使用最好的组件并根据科学应用量身定制设计，SDS 计算机（如 UNIVAC 1108）是 IBM 7090 的自然后代。它们既出售又出租，价格最高可达 250,000 美元。

By 1964 SDS had revenues greater than those of DEC—of course, it had started out with over ten times the capital. In 1969, when SDS had sales of $100 million and after-tax revenues of $10 million, Palevsky sold the company to Xerox for $900 million worth of Xerox stock.⁶⁵ (A few years later Xerox wrote off the division, incurring a loss of $1.3 billion.) Because of its modest beginnings, SDS is often compared to DEC, and its computers are sometimes placed in the class of minicomputers. The 12bit Model 92 was in some ways similar to the PDP-8, but SDS’s main business was the 24-bit 910 and 920—large-scale scientific computers, not as powerful as Control Data’s supercomputers, but much more capable than minicomputers. Another model introduced in 1965, the 940, was explicitly marketed for time-sharing use, and it was also a success. ⁶⁶ Many West Coast time-sharing companies, including the pioneer Tymshare, were based on it. The 940 was used for many pioneering research projects in human-computer interaction and networking in the Palo Alto, California area (before it became known as “Silicon Valley”). In a sense, it was the West Coast counterpart of the DEC PDP-10, a well-engineered, time-sharing system that had its greatest impact in advanced computing research. By using the best components, and by tailoring the design to suit scientific applications, SDS computers, like the UNIVAC 1108, were the natural descendants of the IBM 7090. They were both sold and leased, and prices ranged up to $250,000.

1967 年，SDS 宣布推出一款功能更强大的计算机 Sigma 7，售价约为 100 万美元。67^{帕列夫斯基}在亨茨维尔的关系为他的公司提供了很大的帮助。到 20 世纪 60 年代初，这些设施从陆军转移到了 NASA，在 Wernher von Braun 的领导下，“火箭团队”负责开发将人类送上月球并返回的助推器。IBM 硬件处理了该中心的大部分工作，但 SDS 计算机被安装用于实时模拟和测试火箭的制导系统。Helmut Hoelzer 和 Charles Bradshaw 利用帕列夫斯基在 Bendix 工作时建立的关系，在对最初为此目的订购的 RCA 机器感到失望后，选择安装 SDS 计算机^。68

In 1967 SDS announced a more powerful computer, the Sigma 7, which cost around $1 million.⁶⁷ Palevsky’s Huntsville connections served his company well. By the early 1960s the facilities were transferred from the Army to NASA, where, under the leadership of Wernher von Braun, the “rocket team” was charged with developing the boosters that would take men to the Moon and back. IBM hardware handled the bulk of the Center’s chores, but SDS computers were installed to do real-time simulations and tests of the rockets’ guidance systems. Drawing on a relationship established when Palevsky was working for Bendix, Helmut Hoelzer and Charles Bradshaw chose to install SDS computers after becoming disillusioned with RCA machines they had initially ordered for that purpose.⁶⁸

SDS 的命运随着阿波罗计划而起起伏伏：1969 年，当人类踏上月球时，NASA 却在削减开支，不得不用更少的预算来规划运营。施乐公司以十倍于其收益的价格收购了帕列夫斯基的公司，希望 SDS（现在的 XDS 部门）能够发展壮大。一些记者声称，帕列夫斯基知道他出售的是一家没有前途的公司，但帕列夫斯基在美国对 IBM 反垄断审判的宣誓下表示，他并不这么认为。69^该部门没有发展，施乐公司于 1975 年关闭了 XDS。SDS 没有足够的计划来将其产品扩展到其占据的狭窄市场之外 — — 再次揭示了 IBM 的 System/360 理念的智慧。但施乐公司也必须承担责任。该公司已在加利福尼亚州帕洛阿尔托建立了世界上最好的计算机研究实验室，但未能将其组织的这两个部分整合在一起，更不用说将它们都融入其销售复印机的核心业务中了。

SDS’s fortunes rose and fell with the Apollo program: even as men were walking on the Moon in 1969, NASA was cutting back and having to plan for operations on smaller budgets. Xerox bought Palevsky’s company at a value ten times its earnings, expecting that SDS, now the XDS division, would grow. Some journalists claimed that Palevsky knew he was selling a company with no future, but Palevsky stated, under oath for the United States vs. IBM antitrust trial, that he believed otherwise.⁶⁹ The division did not grow, and Xerox closed XDS in 1975. SDS had no adequate plan for expanding its products beyond the narrow niche it occupied—again revealing the wisdom of IBM’s System/360 philosophy. But Xerox must also shoulder the blame. The company had built up the finest research laboratory for computing in the world, in Palo Alto, California, but it failed to fit these two pieces of its organization together, much less fit both of them into its core business of selling copiers.

软件公司

Software Houses

System/360 如何重新定义计算机行业，其对软件和“服务机构”的影响是最后一个衡量标准。70^成立一家公司，通过购买或租用计算机来为另一家公司的问题提供解决方案，这种想法并不新鲜。第一家可能是 Computer Usage Company，成立于 1955 年，为工业客户开发 IBM 701 和 704 程序。71^大型计算机公司都有自己的内部服务机构，提供相同的服务 — IBM 的服务机构回到了制表机时代，而 Control Data Corporation 的服务业务对公司的财务重要性不亚于其硬件销售。

A final measure of how the System/360 redefined the computer industry was in its effect on software and “service bureaus.”⁷⁰ The idea of forming a company that bought or rented a computer to deliver a solution to another company’s problem was not new. The first may have been Computer Usage Company, founded in 1955, which developed programs for the IBM 701 and 704 for industrial clients.⁷¹ The major computer companies had their own in-house service bureaus that performed the same services—IBM’s went back to the era of tabulators, and Control Data Corporation’s service business was as important financially to the company as its hardware sales.

一家开创性的独立公司是自动数据处理公司，由亨利·陶布于 1949 年在新泽西州帕特森创立，当时名为自动工资单公司。ADP 的核心业务是处理中小型公司的工资单计算。它主要使用 IBM 制表机，即使在 1961 年购买了第一台计算机之后也是如此。第二年，ADP 的收入达到了 100 万美元。72^它对技术采取了保守的态度，使用计算机批量处理打孔卡数据，就像它使用制表机一样。它的第一位销售员弗兰克·劳滕伯格在 1975 年接任首席执行官时延续了陶布保守和以利润为导向的做法。（劳滕伯格后来成为新泽西州的美国参议员。）⁷³

One of the pioneering independent companies was Automatic Data Processing, founded as Automatic Payrolls in 1949 by Henry Taub in Paterson, New Jersey. ADP’s core business was handling payroll calculations for small and medium-sized companies. It primarily used IBM tabulating machinery, even after it acquired its first computer in 1961. The following year ADP’s revenues reached $1 million.⁷² It took a conservative approach to technology, using the computer to process data in batches of punched cards just as it had with its tabulators. Its first salesman, Frank Lautenberg, continued Taub’s conservative and profit-oriented approach when he took over as CEO in 1975. (Lautenberg later became a U.S. senator from New Jersey.)⁷³

计算机科学公司 (Computer Sciences Corporation) 成立于 1959 年，创始人是曾在南加州航空航天业工作的 Fletcher Jones 和 Roy Nutt。如第 3 章所述，CSC 的第一份合同是为霍尼韦尔编写一个商业编程语言 (“FACT”) 编译器。后来，该公司逐渐发展成为一家专注于科学和工程应用的公司，客户包括 NASA-Goddard 太空飞行中心和喷气推进实验室。CSC 还为 20 世纪 60 年代中期销售的大型主机进行基本的系统编程。74^另一家将科学和商业工作结合起来的大公司是 Informatics，由 Walter F. Bauer 于 1963 年创立。

Computer Sciences Corporation was founded in 1959 by Fletcher Jones and Roy Nutt, who had worked in the southern California aerospace industry. As described in chapter 3, CSC’s first contract was to write a compiler for a business programming language (“FACT”) for Honeywell. That evolved into a company that concentrated more on scientific and engineering applications, for customers like the NASA-Goddard Space Flight Center and the Jet Propulsion Laboratory. CSC also did basic systems programming for the large mainframes being sold in the mid-1960s.⁷⁴ Another major company that had a similar mix of scientific and commercial work was Informatics, founded by Walter F. Bauer in 1963.

与小型计算机公司允许第三方 OEM 为特定客户定制系统不同，IBM 的政策是将支持（包括系统分析和编程）包含在硬件的昂贵价格中。1968 年，IBM 同意单独收取这些服务的费用；尽管如此，设置任何 System/360 的复杂性意味着 IBM 必须与客户密切合作，以确保安装顺利进行。“拆分”的决定使原本涓涓细流变成了大量第三方大型机软件和系统公司^。75

In contrast to the minicomputer companies, who let third party OEMs customize a system for specific customers, IBM had a policy of including that support, including systems analysis and programming, into the already substantial price of the hardware. In 1968 IBM agreed to charge for these services separately; still, the complexity of setting up any System/360 meant that IBM had to work closely with its customers to ensure that an installation went well. The decision to “unbundle” turned what had been a trickle into a flood of third-party mainframe software and systems houses.⁷⁵

IBM 360 及其竞争对手等系统的复杂性开辟了新的前景。制造商很难提供使这些计算机有用的所有软件，因为这些机器被设计为同时处理多个任务、支持远程终端、通过网络相互连接，并提供 20 世纪 50 年代后期大型机所不具备的其他功能。商业分时系统的推出开辟了另一条增长途径。许多新的软件公司（如 American Management Systems (AMS)）成立的具体目标是让客户快速掌握这项新的复杂技术。

The complexity of systems like the IBM 360 and its competitors opened up new vistas. Manufacturers were hard-pressed to deliver all the software needed to make these computers useful, because these machines were designed to handle multiple tasks at the same time, support remote terminals, be connected to one another in networks, and deliver other features not present in the mainframes of the late 1950s. The introduction of commercial time-sharing systems opened up still another avenue for growth. Many new software companies, like American Management Systems (AMS), were formed with the specific goal of getting customers up to speed with this new and complex technology.

尽管 IBM 意识到 AMS 这样的公司会对其自身软件团队的收入产生影响，但 IBM 可能还是很高兴有这样一家公司来填补这个空缺。IBM 当时无法提供与其 System/360 硬件一样好的系统和编程软件。最初为 360 设计的操作系统软件交付较晚，交付后运行效果不佳。而原本打算作为 System/360 主要语言的编程语言 PL/I 并没有受到好评。问题来了，IBM 能够如此雄心勃勃地推出新硬件，为什么在为其提供软件方面却失败得如此惨烈？Fred Brooks 写了一本书来回答这个问题，《人月神话》，这本书已成为管理复杂软件项目难度的经典陈述^。76

While mindful of the impact a company like AMS would have on revenues from its own software teams, IBM was probably relieved to have such a company around to step into the breach. IBM was at the time unable to deliver system and programming software that was as good as its System/360 hardware. The original operating system software intended for the 360 was delivered late, and when it was delivered it did not work very well. And the programming language PL/I, intended to be the main language for the System/360, was not well received. The question arose, how could IBM, which could carry off such an ambitious introduction of new hardware, fail so badly in delivering software for it? Fred Brooks wrote a book to answer that question, The Mythical Man-Month, which has become a classic statement of the difficulties of managing complex software projects.⁷⁶

1969 年，IBM 决定将软件定价与硬件分开，此后，IBM 实际上也成为了一家软件公司。有人认为，这一决定是为了阻止传闻中的反垄断行动。（如果真是这样，那它并没有奏效，因为司法部在 IBM 宣布这一决定后的一个月就提起了诉讼。）更准确的说法是，IBM 承认计算机行业已经发生了不可逆转的变化，软件和服务正在成为一个独立的行业^。77

After its decision to unbundle software pricing from hardware in 1969, IBM became, in effect, a software house as well. That decision has been described as an attempt to forestall rumored antitrust action. (If so, it did not work, because the Justice Department filed suit the month after IBM’s announcement.) It is more accurate to say that IBM acknowledged that the computer industry had irrevocably changed, that software and services were becoming a separate industry anyway.⁷⁷

服务和软件提供商的范围不仅涵盖科学和商业，还包括政府和军事承包商。这些公司为专业应用提供了后来被称为“系统集成”的服务。一个例子是 H. Ross Perot 于 1962 年创立的电子数据系统公司 (EDS)。Perot 曾是 IBM 的明星销售员，他曾建议 IBM 设立一个部门，向客户出售计算机时间，而不是计算机本身。当 IBM 拒绝他时，他创办了 EDS。在经历了动荡的开端后，该公司逐渐繁荣起来，在 1965 年国会通过《医疗保险法案》后于 20 世纪 60 年代中期迅速发展。EDS 的大部分业务都是联邦政府的客户^。78

The spectrum of service and software providers not only ran from scientific to commercial, it also included an axis of government and military contractors. These provided what came to be known as “systems integration” for specialized applications. One example was Electronic Data Systems (EDS), founded by H. Ross Perot in 1962. Perot had been a star salesman for IBM, and he had proposed that IBM set up a division that would sell computer time, instead of the computers themselves, to customers. When IBM turned him down he started EDS. After a shaky start, the company prospered, growing rapidly in the mid-1960s after the passage of the Medicare Act by Congress in 1965. Much of EDS’s business was to customers in the federal government.⁷⁸

冷战，尤其是 1957 年人造卫星发射后，导致各种公司开始为国防机构管理系统。这项业务有着深厚的根源，可以追溯到兰德公司及其子公司系统开发公司 (SDC) 的成立，该公司开发防空软件。79^新鲜之处在于，首次出现了希望仅通过承包系统工作来获利的公司，这些公司不像 SDC 那样是联邦资助的国防机构的延伸。位于南加州的 Ramo-Woldridge 可能是其中最成功的一家。它成立于 1953 年，当时 Simon Ramo 和 Dean Woldridge 离开休斯飞机公司，成立了一家专注于机密导弹和太空作战工作的公司。RW 后来被位于俄亥俄州克利夫兰的汽车供应商 Thompson 收购。“锈带”行业与“高科技”的结合可能看起来很糟糕，但结果，TRW 成为这些公司中最赚钱的公司之一。主要原因是汤普森提供了其他系统公司所不具备的制造能力，这使得 TRW 能够作为主要供应商赢得复杂（大部分是机密）太空项目的竞标。20 世纪 60 年代中期，TRW 的净销售额约为 5 亿美元，开始涉足非军事商业工作，并成立了一个开发信用信息数据库的部门。80^然而，该公司仍然专注于军事软件和太空系统。其一名员工巴里·博姆 (Barry Boehm) 帮助创立了“软件工程”学科。TRW 短暂雇用的另一位员工比尔·盖茨 (Bill Gates) 帮助开发了计算机网络软件，该网络管理哥伦比亚河上一系列水坝的水流。（我们将在后面的章节中回顾盖茨在 TRW 的经历以及他后来的职业生涯。）

The Cold War, especially after Sputnik in 1957, led to work for a variety of companies to manage systems for defense agencies. This business had deep roots, going back to the founding of the RAND Corporation and its spin-off, the System Development Corporation (SDC), to develop air defense software.⁷⁹ What was new was that, for the first time, there appeared companies that hoped to make profits only by contracting for systems work, that were not, like SDC, federally funded extensions of a defense agency. Ramo-Woldridge, centered in southern California, was perhaps the most successful of these. It was founded in 1953, when Simon Ramo and Dean Woldridge left Hughes Aircraft to form a company that focused on classified missiles and space operations work. R-W was later acquired by Thompson, an automotive supplier based in Cleveland, Ohio. That marriage of a “rust belt” industry with “high tech” might have seemed a poor one, but the result, TRW, became one of the most profitable of these companies. A major reason was that Thompson supplied a manufacturing capability that the other systems houses lacked, which enabled TRW to win bids for complex (mostly classified) space projects as a prime supplier. In the mid-1960s, with net sales around $500 million, TRW began branching into nonmilitary commercial work, building a division that developed a database of credit information.⁸⁰ The company remained focused on military software and space systems, however. One of its employees, Barry Boehm, helped found the discipline of “software engineering.” Another person TRW employed briefly, Bill Gates, helped develop software for a computer network that managed the flow of water through the series of dams on the Columbia River. (We shall return to Gates’s experience with TRW and his subsequent career in a later chapter.)

除了 TRW 和联邦政府资助的公司（如 SDC 或 MITRE），还有数十家规模较小的公司。他们的共同点是提供软件和支持服务以赚取利润。这些公司大多始于南加州，比如 TRW，通常由航空工程师创立。其中一些公司希望更靠近五角大楼，因此搬到了华盛顿特区，更具体地说，搬到了弗吉尼亚州北部环城公路（1964 年完工）外的开阔农田。这里土地便宜，新修的高速公路使国防部门交通便利。（这些机构和五角大楼一样，主要位于波托马克河的弗吉尼亚州一侧。）⁸¹其中大多数都做得很好，尤其是在罗纳德·里根总统第一任期内从国防合同中获利。大型航空航天和国防公司也开设了部门来服务这个市场。冷战的结束使这些公司陷入混乱，但是他们开创的系统分析具有持久的价值，现在已成为大多数现代行业接受的做法。

Besides TRW and the federally funded companies like SDC or MITRE, there were dozens of smaller fry as well. Their common denominator was that they supplied software and support services for a profit. Most of these began in southern California, like TRW, often founded by aerospace engineers. Some of them, wanting to be closer to the Pentagon, moved to the Washington, D.C., area, more specifically, to the open farmland in northern Virginia just beyond the District’s Beltway (completed in 1964). Here land was cheap, and the new highways made access to the Defense agencies easy. (These agencies, like the Pentagon itself, were mainly on the Virginia side of the Potomac.)⁸¹ Most of them have done very well, especially by profiting from defense contracts during Ronald Reagan’s first term as president. The major aerospace and defense companies also opened up divisions to serve this market. The end of the Cold War has thrown these companies into turmoil, but the systems analysis they pioneered has been of lasting value and is now an accepted practice in most modern industries.

System/360 的最终后果是美国司法部于 1969 年 1 月（约翰逊政府的最后一个工作日）提起的反垄断诉讼。该诉讼拖延了 12 年，各方律师团队需要处理大量文件和工作。（审判文件对历史学家来说是一笔意外之财。）在此期间，IBM 继续盈利并推出新的创新产品；其收入增长了三倍，市场份额保持在 70% 左右。人们不禁想知道，如果没有诉讼，该公司可能会怎么做。诉讼的前提是 IBM 的行为及其对业务的主导地位对“小人物”不利。1982 年 1 月，随着新政府上台，司法部驳回了此案，称其“毫无根据”。82^到1982 年，大型机的地位受到了已经上市几年的个人电脑和刚刚发明的局域网的威胁。尽管 IBM 是个人电脑领域的成功营销者，但正是这些发展（而非司法部）改变了整个行业。如果没有诉讼威胁，IBM 是否会采取更积极的行动来确立其在个人电脑市场的主导地位，这仍是未知数。

A final consequence of the System/360 was, indirectly, the antitrust action filed by the U.S. Justice Department in January 1969, on the last business day of the Johnson Administration. The suit dragged on for twelve years, generating enormous amounts of paper and work for teams of lawyers from all sides. (The documents produced for the trial have been a windfall for historians.) IBM continued to be profitable and to introduce new and innovative products during this time; its revenues tripled and its market share stayed at about 70 percent. One must wonder what the company might have done otherwise. The premise of the action was that IBM’s actions, and its dominance of the business, were detrimental to the “dwarfs.” In January 1982, with a new administration in power, the Justice Department dismissed the case, stating that it was “without merit.”⁸² By 1982 the place of the mainframe was being threatened by the personal computer, which had already been on the market for a few years, and by local-area networking, just invented. These developments, not the Justice Department, restructured the industry, in spite of IBM’s role as a successful marketer of personal computers. Whether IBM would have acted more aggressively in establishing its dominance of the PC market had there been no threat of litigation remains unanswered.

一群人的命运

The Fate of the BUNCH

司法部诉讼暗示，BUNCH 的生存正受到 IBM 政策的威胁。讽刺的是，BUNCH 的每一位成员都面临着与 IBM 无关的悲惨命运。

The Justice Department suit implied that the BUNCH’s very existence was being threatened by IBM’s policies. Ironically, each of the BUNCH faced a depressing fate that had little to do with IBM.

1986 年，Burroughs 和 UNIVAC 合并为 Unisys，该公司一度成为第二大计算机公司。从 Eckert and Mauchly 到 Remington Rand，再到 Sperry，再到 Burroughs，UNIVAC 这个名字在某个时候被遗忘了。到 1986 年，很少有人记得“UNIVAC”曾经是“计算机”的同义词，就像“Scotch”胶带或“Thermos”瓶一样。随意放弃这个受人尊敬的名字或许象征着 Unisys 的困境；几年后，它开始亏损并跌至较低级别。它大幅裁员，经过一些痛苦的重组后开始显示一些利润。

In 1986 Burroughs and UNIVAC merged into a company called Unisys, which briefly became the second-largest computer company. In its travels from Eckert and Mauchly, to Remington Rand, to Sperry, to Burroughs, the name UNIVAC was somewhere dropped. By 1986 few remembered that “UNIVAC” was once synonymous with “computer,” like “Scotch” tape or “Thermos” bottle. The casual abandonment of this venerated name was perhaps symbolic of the troubles of Unisys; with a few years it began suffering losses and fell to the lower ranks. It cut employment drastically, and after some painful restructuring began to show some profits.

20 世纪 80 年代，NCR 大胆尝试采用基于廉价微处理器和非专有 UNIX 操作系统的新架构。它是首批这样做的大型系统公司之一。NCR 还率先构建了能够通过小型微处理器子单元集群实现大型机性能的系统，这是许多其他公司一直追求但未获成功的圣杯。但其创新文化使该公司成为收购目标。1991 年，现已解除管制的 AT&T 为在大型商业系统中占据竞争地位，通过恶意收购收购了 NCR。与 Burroughs-Univac 合并一样，这次收购也令人失望。AT&T 向 NCR 员工承诺，它将保留这家计算机公司的管理结构、文化，甚至首字母缩写（意思是“网络计算资源”，而不是“国家收银机”）。但几年后，当 SUN 和 Silicon Graphics 等公司抢先推出此类产品进入市场时，AT&T 违背了这三个承诺。 AT&T 于 1996 年将 NCR 剥离出来，成为一家独立公司。

In the 1980s NCR made a brave attempt to adopt the new architectures based on cheap microprocessors and the nonproprietary UNIX operating system. It was one of the first large system companies to do so. NCR also pioneered in building systems that gave mainframe performance from clusters of smaller, microprocessor-based subunits—a Holy Grail that many others had sought with little success. But its innovative culture made the company a takeover target. In 1991, a now-deregulated AT&T, seeking to vault into a competitive position in large commercial systems, bought NCR in a hostile takeover. Like the Burroughs-Univac combination, this was also a disappointment. AT&T promised NCR employees that it would preserve the computer company’s management structure, culture, and even the initials (to mean “Networked Computing Resources” instead of “National Cash Register”). But a few years later AT&T broke all three promises when companies like SUN and Silicon Graphics beat them to market with these kinds of products. AT&T spun off NCR as an independent company in 1996.

霍尼韦尔与日本电气公司 (NEC) 结盟，制造与 IBM 兼容的大型机。自 1970 年代以来，它还与法国公司 Machines Bull 和意大利公司 Olivetti 结盟。从 1986 年开始，霍尼韦尔开始退出大型机业务，第二年将其完全转交给 Bull，NEC 是小合作伙伴。83 霍尼韦尔^继续向美国军方市场供应国产产品，并与 Sperry 一起成为军用和民用专用航空计算机的领先供应商——随着新一代飞机采用“电传操纵”控制系统，这一领域正在不断发展。20 世纪 80 年代中期，霍尼韦尔根据军方合同开发了一套称为 VHSIC（超高速集成电路）的专用芯片，这种芯片具有抗辐射性。但与二十年前的情况不同，集成电路的军方合同并没有很好地转化为商业产品^。84

Honeywell allied itself with the Nippon Electric Company (NEC) to build its mainframes, which were IBM compatible. It had also been allied since the 1970s with the French company Machines Bull and the Italian company Olivetti. Beginning in 1986, Honeywell began a retreat out of the mainframe business and the next year turned it completely over to Bull, with NEC a minor partner.⁸³ Honeywell continued supplying the U.S. military market with domestic products, and along with Sperry became a leading supplier of specialized aerospace computers, military and civilian—a growing field as new-generation aircraft adopted “fly-bywire” controls. In the mid-1980s Honeywell developed, under military contract, a set of specialized chips called VHSIC (Very High Speed Integrated Circuits), which were resistant to radiation. But unlike the situation two decades earlier, military contracts for integrated circuits did not lead nicely to commercial products.⁸⁴

Control Data 有着一段不寻常的历史。它发展了一项健康的业务，为竞争对手的计算机制造磁带驱动器和打印机，并进入了服务业。1968 年，随着其股票在繁荣时期达到顶峰，它利用这些股票收购了巴尔的摩金融公司 Commercial Credit——一家比 CDC 大很多倍的公司。此次收购为 CDC 提供了资金来源，为其多元化提供资金。一些观察家指责 CDC 在接下来的二十年里榨干了 Commercial Credit 的资产，耗尽了它的活力，这是 1980 年代杠杆收购的预兆。85^与大多数起诉 IBM 的公司不同，Control Data 在 1973 年达成了有利的和解。这导致 IBM 将自己的服务局转移到^CDC。86

Control Data had an unusual history. It developed a healthy business of manufacturing tape drives and printers for competitors’ computers, and it entered the service business as well. In 1968, with its stock riding the crest of the go-go years, it used that stock to acquire the Baltimore finance company Commercial Credit—a company many times larger than CDC. The acquisition gave CDC a source of funds to finance its diversification. Some observers charge that CDC milked the assets of Commercial Credit and drained it of its vitality over the next two decades, a foreshadowing of the leveraged buyouts of the 1980s.⁸⁵ Unlike most of the companies that brought suit against IBM, Control Data achieved a favorable settlement in 1973. That resulted in IBM’s transferring its own Service Bureau to CDC.⁸⁶

这些胜利让 CDC 的创始人兼董事长比尔·诺里斯 (Bill Norris) 看起来像一只狡猾的狐狸，但我们现在知道，诺里斯犯了一个不可饶恕的错误，那就是忽视了技术的发展步伐。87 ^CDC的成功来自其产品的卓越性能，尤其是超级计算机——CDC 首创的一类机器。诺里斯在法庭上获胜的能力或玩弄膨胀的股票的能力是无可替代的。CDC 从未真正摆脱西摩·克雷的离开。1972 年，克雷在威斯康星州奇珀瓦福尔斯的家旁边建立了克雷研究公司，并着手重塑 CDC 早期的精神。CRAY-1 于 1976 年推出，并开启了一系列成功的超级计算机。CDC 继续推出超级计算机，但没有一台能与西摩·克雷实验室的产品相媲美。

These victories made Bill Norris, CDC’s founder and chairman, look like a wily fox, but we now know that Norris made the unforgivable error of taking his eye off the advancing pace of technology.⁸⁷ CDC’s success came from the superior performance of its products, especially supercomputers—a class of machines that CDC pioneered. Norris’s ability to win in the courtroom or play with inflated stock was no substitute. CDC never really survived Seymour Cray’s leaving. In 1972 Cray founded Cray Research, with a laboratory next to his house in Chippewa Falls, Wisconsin, and set out to recreate the spirit of CDC’s early days. The CRAY-1 was introduced in 1976 and inaugurated a series of successful supercomputers. CDC continued to introduce supercomputers, but none could match the products from Seymour Cray’s laboratory.

更令人心碎的是 CDC 的 PLATO 的失败，PLATO 是一个基于图形的交互式系统，旨在为从幼儿园到小学的各个层次的教育和培训提供服务（图 5.5）。它承诺，无论是专家还是外行，都可以轻松直接地访问来自世界各地图书馆和档案馆的信息。CDC 斥资数百万美元开发 PLATO，并于 20 世纪 70 年代中期在伊利诺伊大学进行了大规模的试点安装。88^但它最终还是失败了。原因很复杂。PLATO 需要在 CDC 中央主机上运行，终端价格昂贵，而且 PLATO 可能太超前了。1994 年，通过互联网和使用万维网，PLATO 的大部分预测都成真了。（请注意，联邦政府支付了这些系统的大部分研发费用。）到那时，CDC 已经无法从 PLATO 中获得任何收益了。 20 世纪 80 年代中期，公司开始出现巨额亏损，1986 年比尔·诺里斯辞职。CDC 幸存了下来，但只是一家专业硬件和服务供应商，主要面向不断萎缩的军事市场。

Even more heartbreaking was the failure of CDC’s PLATO, an interactive, graphics-based system intended for education and training at all levels, from kindergarten on up (figure 5.5). It promised, for the expert and lay-person alike, easy and direct access to information from libraries and archives worldwide. CDC spent millions developing PLATO and had a large pilot installation operating at the University of Illinois by the mid-1970s.⁸⁸ But ultimately it failed. The reasons are complex. PLATO required a central CDC mainframe to run on, the terminals were expensive, and PLATO may have been too far ahead of its time. In 1994 most of the predictions for PLATO came true, via the Internet and using a system called the World Wide Web. (Note that the federal government paid most of the R&D costs of these systems.) By then it was too late for CDC to reap any benefits from PLATO. The company began losing large amounts of money in the mid-1980s, and in 1986 Bill Norris resigned. CDC survived, but only as a supplier of specialized hardware and services, mainly to an ever-shrinking military market.

结论

Conclusion

约翰·布鲁克斯的“辉煌岁月”如今已成为遥远的记忆。相比之下，苹果、微软和其他公司在 20 世纪 80 年代和 90 年代的故事让早期的故事显得平淡无奇。人们记得当时飞速发展的金融业务，但他们忘记了那些年为后来的计算机行业转型奠定了基础。这些基础包括使用集成电路构建大型系统、使用磁盘文件的大型数据存储，以及最重要的是用高级语言编写的复杂软件。独立软件和系统公司以及插件兼容制造商的崛起也预示着软件公司将成为计算领域平等甚至占主导地位的合作伙伴，而计算机架构的克隆也将变得普遍。最后，华尔街意识到计算机、半导体和软件应该像雷丁铁路或美国钢铁公司一样受到关注。

John Brooks’s “go-go years” are now a distant memory. The stories of Apple, Microsoft, and other companies from the 1980s and 1990s make those of an earlier era seem tame by comparison. People remember the high-flying financial doings, but they forget that those were the years when the foundation was laid for later transformations of the computer industry. That foundation included building large systems using integrated circuits, large data stores using disk files, and above all complex software written in high-level languages. The rise of independent software and systems houses, as well as plug-compatible manufacturers, also foreshadowed a time when software companies would become equal if not dominant partners in computing, and when clones of computer architectures also became common. Finally, it was a time when Wall Street learned that computers, semiconductors, and software deserved as much attention as the Reading Railroad or United States Steel.

图 5.5

CDC 的 PLATO 系统。（顶部）PLATO 的一个用途是存储和检索工程图纸和数据。（中间）另一个被广泛宣传的用途是教育。（底部）残疾人正在使用 PLATO 终端（请注意靠在桌子上的支架）。控制数据负责人 William Norris 曾广泛撰写和演讲，阐述计算机向普通人开放时的社会效益。这张照片无意中揭示了 PLATO 最终失败的原因。背景是 Radio Shack 的早期个人计算机模型。与 PLATO 相比，它非常原始，但最终个人计算机成为向家庭提供计算和电信服务的基础，而成本仅为 PLATO 的一小部分。（来源：明尼苏达大学查尔斯·巴贝奇研究所。）

Figure 5.5

CDC’s PLATO System. (top) One use for PLATO was to store and retrieve engineering drawings and data. (middle) Another use, one that was widely publicized, was for education. (bottom) A PLATO terminal being used by a handicapped person (note the brace leaning against the desk). William Norris, the head of Control Data, wrote and spoke extensively on the social benefits of computing when made available to lay persons. The photograph inadvertently reveals why PLATO ultimately failed. In the background is an early model of a personal computer from Radio Shack. It is very primitive in comparison to PLATO, but eventually personal computers became the basis for delivering computing and telecommunications to the home, at a fraction of the cost of PLATO. (Source : Charles Babbage Institute, University of Minnesota.)

芯片及其影响，1965-1975

The Chip and Its Impact, 1965–1975

正如 IBM System/360 改变了大型计算机一样，20 世纪 60 年代末的一系列新机器也改变了小型计算机。起初，这两个计算部分独立运作，但在 20 世纪 70 年代，它们开始合并。这些变化的背后是一项名为集成电路的发明，现在被普遍称为“芯片”。

Just as the IBM System/360 transformed mainframe computing, so did a series of new machines transform minicomputing in the late 1960s. At first these two computing segments operated independently, but during the 1970s they began to coalesce. Behind these changes was an invention called the integrated circuit, now known universally as “the chip.”

小型计算机（如 PDP-8）并未对大型计算机业务构成威胁；它们开拓了尚未开发的市场，并与大型计算机共存。有人认为，用一组小型计算机完成大型计算机的工作是可能的，而且成本要低得多。大型计算机销售人员引用“格罗施定律”，认为这种诱人的想法违背了计算机有利于大型系统的基本特征。该定律以计算机行业著名人物 Herb Grosch（图 6.1 ）的名字命名，它指出，两倍大的计算机系统（即花费两倍的^钱）可为您带来不是两倍而是四倍的计算能力。如果您购买两台小型计算机，其计算能力是一台计算机的两倍，那么您将不如用这笔钱购买一台大型计算机。1

Minicomputers such as the PDP-8 did not threaten mainframe business; they exploited an untapped market and lived in symbiosis with their large cousins. Some thought it might be possible to do a mainframe’s work with an ensemble of minis, at far lower cost. Mainframe salesmen, citing “Grosch’s Law,” argued that this tempting idea went against a fundamental characteristic of computers that favored large systems. Named for Herb Grosch (figure 6.1), a colorful figure in the computer business, this law stated that a computer system that was twice as big (i.e., that cost you twice as much money) got you not twice but four times as much computing power. If you bought two small computers, giving you two times the power of a single one, you would not do as well as you would if you used the money to buy a single larger computer.¹

相信该定律的人列举了几个理由。那个时代的计算机使用磁芯进行存储。磁芯本身很便宜，但读取、写入和擦除磁芯上信息所需的支持电路却很昂贵。而且无论内存容量是大还是小，都需要一定数量的支持电路。这使得小内存的每比特成本高于大内存，因此选择后者更为经济，并附带一个大型处理器系统来利用它。最令人信服的原因是，没有人真正知道如何将小型计算机连接在一起，并使整个系统协调一致。这就像试图用一队双翼飞机而不是一架巨型喷气式飞机将乘客送过大西洋一样。最终，随着半导体存储器和新网络架构的出现，这两个障碍都会消失。到那时——大约在 20 世纪 80 年代中期——小型计算机本身已经被基于微处理器的工作站所取代。²但随着 20 世纪 60 年代末小型计算机功能越来越强大，它们开始慢慢渗透到大型计算机领域，同时开辟了新的应用领域。格罗施定律虽然成立，但已不再适用。

Believers in that law cited several reasons for it. Computers of that era used magnetic cores for storage. The cores themselves were cheap, but the support circuitry needed to read, write, and erase information on them was expensive. And a certain amount of that circuitry was required whether a memory capacity was large or small. That made the cost per bit higher for small memories than for large, so it was more economical to choose the latter, with an accompanying large processor system to take advantage of it. The most compelling reason was that no one really knew how to link small computers to one another and get coordinated performance out of the ensemble. It would have been like trying to fly passengers across the Atlantic with an armada of biplanes instead of a single jumbo jet. Eventually both barriers would fall, with the advent of semiconductor memory and new network architectures. By the time that happened—around the mid 1980s—the minicomputer itself had been replaced by a microprocessor-based workstation.² But as minicomputers had grown more and more capable through the late 1960s, they had slowly begun a penetration into mainframe territory while opening up new areas of application. Grosch’s Law held, but it no longer ruled.

图 6.1

Herbert Grosch，大约1955 年。（来源： Herbert Grosch。）

Figure 6.1

Herbert Grosch, ca. 1955. (Source : Herbert Grosch.)

推动小型计算机发展的动力是其基本电路的改进，这种改进始于 1959 年的集成电路 (IC)。IC 或芯片取代了计算机处理单元中的晶体管、电阻器和其他分立电路；它还取代了内存单元的核心。芯片对社会的影响一直是无休止的讨论和分析的主题。本章也将提供分析，认识到芯片是一种进化发展，其起源可以追溯到第一台电子数字计算机的电路设计，甚至可能更早。

The force that drove the minicomputer was an improvement in its basic circuits, which began with the integrated circuit (IC) in 1959. The IC, or chip, replaced transistors, resistors, and other discrete circuits in the processing units of computers; it also replaced cores for the memory units. The chip’s impact on society has been the subject of endless discussion and analysis. This chapter, too, will offer an analysis, recognizing that the chip was an evolutionary development whose origins go back to the circuit designs of the first electronic digital computers, and perhaps before that.

冯·诺依曼体系结构用四个基本功能单元来描述计算机：内存、处理器、输入和输出。在此级别之下是功能构建块，它们执行逻辑运算“与”、“或”、“非”、“异或”和其他一些运算。在此之下是电路，每个电路都需要几个（最多约十几个）电子工程师熟悉的元件：电子管（后来是晶体管）、电阻器、电容器、电感器和电线。在 20 世纪 40 年代，任何制造计算机的人都必须从这个级别进行设计。但随着计算机设计成为一门独立的学科，它开始在更高的级别上进行设计，即在二进制数字集上运行逻辑函数的级别。因此产生了将组件组装成模块的想法，这些模块的电气特性是标准化的，并执行逻辑功能。使用标准化模块不仅简化了计算机设计，还简化了测试和维护，这两项工作在易碎的真空管时代都至关重要。

The von Neumann architecture described a computer in terms of its four basic functional units—memory, processor, input, and output. Below that level were the functional building blocks, which carried out the logical operations “AND,” “OR,” “NOT,” “EXCLUSIVE OR,” and a few others. Below that were circuits that each required a few—up to about a dozen—components that electrical engineers were familiar with: tubes (later transistors), resistors, capacitors, inductors, and wire. In the 1940s anyone who built a computer had to design from that level. But as computer design emerged as a discipline of its own, it did so at a higher level, the level of the logical functions operating on sets of binary digits. Thus arose the idea of assembling components into modules whose electrical properties were standardized, and which carried out a logical function. Using standardized modules simplified not only computer design but also testing and maintenance, both crucial activities in the era of fragile vacuum tubes.

J. Presper Eckert 率先在 ENIAC 中使用模块处理一组十进制数字，并在 UNIVAC 中使用模块处理二进制编码的数字，这是一项关键且经常被忽视的发明，它确保了这两台计算机的长期实用性，而当时其他计算机一次工作的时间很少超过一小时。3^当IBM 携其 701 型进入该行业时，它还开发了电路模块 — — 需要两千多种不同的模块。对于其晶体管机器，它开发了一种紧凑而多功能的“标准模块化系统”，从而减少了不同类型的数量。4数字设备公司成立第一年推出的第^一批也是唯一一批产品是逻辑模块，而其 PDP-8 的成功取决于由安装在小型电路板上的分立器件组成的“倒装芯片”模块。

J. Presper Eckert pioneered in using modules in the ENIAC to handle a group of decimal digits, and in the UNIVAC to handle digits coded in binary, a key and often overlooked invention that ensured the long-term usefulness of those two computers, at a time when other computers seldom worked more than an hour at a time.³ When IBM entered the business with its Model 701, it also developed circuit modules—over two thousand different ones were required. For its transistorized machines it developed a compact and versatile “Standard Modular System” that reduced the number of different types.⁴ Digital Equipment Corporation’s first, and only, products for its first year of existence were logic modules, and the success of its PDP-8 depended on “flip-chip” modules that consisted of discrete devices mounted on small circuit boards.

1959 年，德州仪器公司的杰克·基尔比 (Jack Kilby) 和仙童半导体公司的罗伯特·诺伊斯 (Robert Noyce) 申请了在单个电路上结合多种操作的设备的专利。他们的发明最初被称为“Micrologic”，仙童半导体公司后来将其命名为“集成电路”，这只是沿着这条道路迈出的又一步。5^基尔比和诺伊斯都知道，当时的主流观点认为，现有的小型化和互连设备方法（包括上述方法）是不够的。美国空军对新事物提出了巨大的推动力，他们需要在弹道导弹和飞机上配备更复杂的电子设备，而这两者都对重量、功耗和空间有严格的要求。（仔细研究空军的需求就会发现，他们最关心的是可靠性，而不是尺寸。6 ^）民用电子市场也想要一些东西，主要关注手工接线计算机电路的成本和错误。对于 PDP-8 的生产，自动绕线机连接倒装芯片模块。用戈登·贝尔的话来说，这样就省去了“一整层楼的小女士们来接线电脑”的麻烦，尽管制造一台电脑仍然是劳动密集型的。7^简而言之，“技术界的很大一部分人都在寻找问题的解决方案，因为很明显，一个现成的市场正在等待成功的发明家。” ⁸

Patents for devices that combined more than one operation on a single circuit were filed in 1959 by Jack Kilby of Texas Instruments and Robert Noyce of Fairchild Semiconductor. Their invention, dubbed at first “Micrologic,” then the “Integrated Circuit” by Fairchild, was simply another step along this path.⁵ Both Kilby and Noyce were aware of the prevailing opinion that existing methods of miniaturization and of interconnecting devices, including those described above, were inadequate. A substantial push for something new had come from the U.S. Air Force, which needed ever more sophisticated electronic equipment onboard ballistic missiles and airplanes, both of which had stringent weight, power consumption, and space requirements. (A closer look at the Air Force’s needs reveals that reliability, more than size, was foremost on its mind.⁶) The civilian electronics market, which wanted something as well, was primarily concerned with the costs and errors that accompanied the wiring of computer circuits by hand. For the PDP-8’s production, automatic wire-wrap machines connected the flip-chip modules. That eliminated, in Gordon Bell’s words, “a whole floor full of little ladies wiring computers,” although building a computer was still labor-intensive.⁷ In short, “[a] large segment of the technical community was on the lookout for a solution of the problem because it was clear that a ready market awaited the successful inventor.”⁸

在显微镜下观察现代集成电路，它看起来就像一个大型未来大都市的规划图。在描述芯片设计和布局时，将其与建筑设计或城市规划进行类比是恰当的。芯片管理电力、信号和热量的流动，就像城市管理人流、货物和能源的流动一样。一个更具启发性的类比是印刷术，尤其是用照相方法进行的印刷术。现代集成电路价格低廉的原因与平装书价格低廉的原因相同——材料便宜，可以批量生产。它们像缩微胶片一样，在小容量中存储大量信息。从历史上看，印刷术、照相术和微电子学之间的关系一直很密切。

Modern integrated circuits, when examined under a microscope, look like the plan of a large, futuristic metropolis. The analogy with architectural design or city planning is appropriate when describing chip design and layout. Chips manage the flow of power, signals, and heat just as cities handle the flow of people, goods, and energy. A more illuminating analogy is with printing, especially printing by photographic methods. Modern integrated circuits are inexpensive for the same reasons that a paperback book is inexpensive—the material is cheap and they can be mass produced. They store a lot of information in a small volume just as microfilm does. Historically, the relationship between printing, photography, and microelectronics has been a close one.

类似数字设备公司的倒装芯片之类的模块通过在覆盖有铜或其他导体的塑料板上蚀刻图案来连接组件；然后将电路板浸入溶剂中，除去蚀刻图案保护的导体以外的所有导体。第二次世界大战期间，这种技术在几个地方率先采用，包括位于威斯康星州密尔沃基的 Globe-Union 公司的中央实验室部门，在那里生产了盟军使用的火炮引信的电路。其他工作由华盛顿特区的国家标准局完成⁹其中一些工作基于保罗·艾斯勒 (Paul Eisler) 获得的专利，保罗·艾斯勒是一名在战争期间在英国工作的奥地利难民，艾斯勒声称他的印刷电路被用于战争中最著名的微型电子产品——近炸引信，尽管其他人对这一说法提出异议。¹⁰在他 1948 年授予的专利中，艾斯勒将他的发明描述为“一种基于印刷导电金属图像的工艺”。¹¹战后，美国陆军通信兵采用了后来被称为“印刷电路”的技术，并对其进行了进一步开发。陆军将其称为“自动组装”，以强调生产而不是小型化。12^它是印刷电路的前身，为消费者和军用市场所熟知，至今仍在使用^。13

Modules like Digital Equipment Corporation’s flip chips interconnected components by etching a pattern on a plastic board covered with copper or some other conductor; the board was then dipped into a solvent that removed all the conductor except what was protected by the etched pattern. This technique was pioneered during the Second World War in several places, including the Centrallab Division of the Globe-Union Company in Milwaukee, Wisconsin, where circuits were produced for an artillery fuze used by allied forces. Other work was done at the National Bureau of Standards in Washington, D.C.⁹ Some of this work was based on patents taken out by Paul Eisler, an Austrian refugee who worked in England during the war, Eisler claims his printed circuits were used in the war’s most famous example of miniaturized electronics, the Proximity Fuze, although others dispute that claim.¹⁰ In his patent granted in 1948, Eisler describes his invention as “a process based on the printing of a representation of the conductive metal.”¹¹ After the war the “printed circuit,” as it became known, was adopted by the U.S. Army’s Signal Corps for further development. The Army called it “Auto-Sembly” to emphasize production rather than its miniaturization.¹² It was the ancestor of printed circuits, familiar to both the consumer and military markets, and still in use.¹³

整个 20 世纪 50 年代，美国武装部队都在寻求解决互连问题的方法，认为这是提高可靠性的一种可能方法。可靠性是美国空军特别关注的问题，他们曾因数百万美元的火箭发射失败而感到尴尬，后来发现这些失败是由最多花费几美元的故障部件引起的。空军针对民兵弹道导弹计划的这一问题发起了直接攻势，建立了一套深入部件制造商生产线的正式程序。

Throughout the 1950s, the U.S. armed services pressed for a solution to the interconnection problem, seeing it as a possible way to increase reliability. Reliability was of special concern to the U.S. Air Force, which had found itself embarrassed by failures of multimillion dollar rocket launches, failures later found to have been caused by a faulty component that cost at most a few dollars. The Air Force mounted a direct attack on this problem for the Minuteman ballistic missile program, setting up a formal procedure that penetrated deep into the production lines of the components’ manufacturers.

与此同时，空军启动了一项雄心勃勃的计划，称之为“分子电子学”，其目标是开发由物质构成的新设备，这些物质的单个分子可以进行开关。具体如何实现尚不清楚，但空军在 1959 年 4 月将一份价值 200 万美元的开发合同授予威斯汀豪斯公司，让其进行尝试——这距离集成电路发明仅几个月。14^后来，威斯汀豪斯公司又获得了 260 万美元。但这个想法最终没有得到实现。授予合同两年后，空军和威斯汀豪斯公司报告了重大进展，但媒体报道称“美国空军对冲分子电子学赌注”，称使用集成电路是缩小机载电子设备尺寸和复杂性所需的“临时措施”。15 ^“分子电子学”一词从随后的报道中悄然消失。

At the same time it inaugurated an ambitious program it called “molecular electronics,” whose goal was to develop new devices made of substances whose individual molecules did the switching. Just how that would be done was unspecified, but the Air Force awarded a $2 million development contract to Westinghouse in April 1959—within months of the invention of the IC—to try.¹⁴ Later on Westinghouse received another $2.6 million. The idea never really went anywhere. Two years after awarding the contract, the Air Force and Westinghouse reported substantial progress, but the press, reporting that the “USAF Hedges Molectronics Bets,” called the use of ICs an “interim step” needed to reduce the size and complexity of airborne electronics.¹⁵ The term “molecular electronics” quietly vanished from subsequent reports.

美国空军对民兵弹道导弹零部件可靠性的追求对电子行业产生了更大的影响，因为它确实在可靠性方面取得了突破。供应商引入了“洁净室”，工人们在里面穿着工作服以防止灰尘进入他们正在处理的材料。这种洁净室是桑迪亚国家实验室在 20 世纪 60 年代初为核武器组装而发明的，通过从天花板到地板的恒定超滤空气流来冲洗。16^最终，该行业将建造比医院干净许多倍的制造室或“晶圆厂”。它们可以在精确调节的温度和压力下，将材料中的杂质控制在几乎原子级的水平。电子行业开发了这些技术来为民兵制造晶体管。这种文化扎根了。

The Air Force’s push for higher reliability of parts for the Minuteman ballistic missile had a greater impact on the electronics industry because it did achieve a breakthrough in reliability. Suppliers introduced “clean rooms,” where workers wore gowns to keep dust away from the materials they were working with. Invented at the Sandia National Laboratories in the early 1960s for atomic weapons assembly, such rooms were washed by a constant flow of ultra-filtered air flowing from the ceiling to the floor.¹⁶ Eventually the industry would build fabrication rooms, or “fabs,” that were many times cleaner than a hospital. They would control the impurities of materials almost to an atom-by-atom level, at temperatures and pressures regulated precisely. The electronics industry developed these techniques to make transistors for Minuteman. The culture took root.

在民兵使用的每一个电子元件的生产过程中，都会保存一份日志，上面详细记录了谁对这个元件做了什么操作。如果一个元件在后续测试中不合格，即使是几个月后进行的测试，人们也可以回溯并找出它出现在哪里。如果故障是由于生产运行失误造成的，那么使用那次运行中元件的每个系统都会被识别出来并从服务中移除。不能或不愿遵守这些程序的供应商将被淘汰。17^通过测试的供应商还发现了额外的好处：他们可以把自己的元件作为符合“民兵高可靠性”标准的元件在其他地方销售，并比竞争对手生产的元件收取更高的价格。最终，空军要求的故障率降低约一百倍，也逐渐被商业世界接受为标准。18 与^格雷欣定律相反，高质量将低质量商品赶出了市场。

At every step of the production of every electronic component used in Minuteman, a log was kept that spelled out exactly what was done to the part, and by whom. If a part failed a subsequent test, even a test performed months later, one could go back and find out where it had been. If the failure was due to a faulty production run, then every system that used parts from that run could be identified and removed from service. Suppliers who could not or would not follow these procedures were dropped.¹⁷ Those who passed the test found an additional benefit: they could market their components elsewhere as meeting the “Minuteman Hi-Rel” standard, charging a premium over components produced by their competitors. Eventually the estimated hundred-fold reduction of failure rates demanded by the Air Force came to be accepted as the norm for the commercial world as well.¹⁸ In a reverse of Gresham’s Law, high-quality drove low-quality goods from the market.

该计划的代价非常高昂。每枚民兵导弹的成本在 300 万到 1000 万美元之间，其中高达 40% 的成本用于电子设备。19 ^Hi -Rel 计划的重点仍然是分立元件，尽管后来洁净室生产技术转移到了 IC 生产。无论民兵导弹多么成功，Hi-Rel 计划并没有自动带来商业市场（更不用说消费者市场）的进步^。20

This program came at a steep price. Each Minuteman in a silo cost between $3 and $10 million, of which up to 40 percent was for the electronics.¹⁹ And the Hi-Rel program’s emphasis remained on discrete components, although the clean-room production techniques were later transferred to IC production. However successful it was for the Minuteman, the Hi-Rel program did not automatically lead to advances in commercial, much less consumer, markets.²⁰

集成电路的发明

The Invention of the Integrated Circuit

20 世纪 60 年代初，美国空军开始研发改进型“民兵”导弹，这种导弹的制导要求远远超出了现有导弹计算机的处理能力。主要出于政治原因，“那些希望洲际弹道导弹具备其他能力的人无法重新开始研制一种全新的导弹。相反，他们不得不设法将自己想要的东西融入到一代又一代的“民兵”导弹中。” ²¹到 20 世纪 60 年代中期，对“民兵”导弹制导系统的重新设计促使美国空军大量采购这种新发明的 IC，正是这些采购推动了 IC 进入商业市场。

In the early 1960s the Air Force initiated the development of an improved Minuteman, one whose guidance requirements were far greater than the existing missile’s computer could handle. For mainly political reasons, “those who wished other capabilities from ICBMs [intercontinental ballistic missiles] were unable to start afresh with an entirely new missile. Instead, they had to seek to build what they wanted into successive generations of Minuteman.”²¹ The reengineering of Minuteman’s guidance system led, by the mid-1960s, to massive Air Force purchases for the newly invented IC, and it was those purchases that helped propel the IC into the commercial marketplace.

在讨论这些事件之前，有必要先了解一下 IC 的发明背景。尽管军方和 NASA 是 IC 的重要客户，但他们对 IC 的发明影响甚微。

Before discussing those events, it is worth looking at the circumstances surrounding the IC’s invention. As important as the military and NASA were as customers for the IC, they had little to do with shaping its invention.

1947 年，杰克·基尔比从伊利诺伊大学电气工程专业毕业后，进入密尔沃基的 Centrallab 公司工作，该公司是印刷电路和微型化领域的行业领导者。起初，他从事印刷电路设计工作；后来，他参与让公司使用锗晶体管生产产品。“到 1957 年……很明显，很快就需要大量支出。军事市场代表着一个重大机遇，但需要硅器件。……扩散晶体管的优势逐渐显现，其开发也需要超出 Centrallab 能力的支出。……我决定离开公司。” ²²第二年，他加入了达拉斯的德州仪器公司，该公司在业界因率先从锗晶体管转向硅晶体管而闻名。“我的职责没有明确规定，但大家都知道我将在微型化领域工作。” ²³德州仪器 (TI) 是众多认识到此类设备潜在军用和民用市场的公司之一。但如何打造它们呢？

After graduating from the University of Illinois with a degree in Electrical Engineering in 1947, Jack Kilby took a job at Centrallab in Milwaukee—the industrial leader in printed circuits and miniaturization. At first he worked on printed circuit design; later he became involved in getting the company to make products using germanium transistors. “By 1957 ... it was clear that major expenditures would soon be required. The military market represented a major opportunity, but required silicon devices. ... The advantages of the diffused transistor were becoming apparent, and its development would also have required expenditures beyond the capabilities of Centrallab. ... I decided to leave the company.”²² The following year he joined Texas Instruments in Dallas, already known in the industry for having pioneered the shift from germanium to silicon transistors. “My duties were not precisely defined, but it was understood that I would work in the general area of microminiaturization.”²³ Texas Instruments (TI) was one among many companies that recognized the potential market, both military and civilian, for such devices. But how to build them?

杰克·基尔比身材高大，为人谦逊，他安静的举止反映了人们通常认为中西部人解决问题的务实态度。他出生于密苏里州杰斐逊市，在堪萨斯州大本德的农业和油井供应小镇长大，大本德因阿肯色河从落基山脉流出后向南拐弯而得名。他的父亲是当地一家电力公司的工程师。24^他回忆起从父亲那里学到的东西，在工程解决方案中，某物的成本与其他变量一样重要^。25

Jack Kilby is a tall, modest man whose quiet manner reflects the practical approach to problems people often associate with Midwesterners. He was born in Jefferson City, Missouri, and grew up in the farming and oil-well supply town of Great Bend, Kansas, named after the southern turn that the Arkansas River takes after coming out of the Rockies. His father was an engineer for a local electrical utility.²⁴ He recalls learning from his father that the cost of something was as important a variable in an engineering solution as any other.²⁵

1958 年，正如 TI 和其他地方的其他人所做的那样，Kilby 研究了微型化，并对当时正在进行的各种政府资助项目进行了评估。这些项目中有一个 TI 已经参与的项目，称为“微模块”，涉及将组件放置在陶瓷晶片上。26 ^Kilby认为这种方法不具成本效益（尽管 IBM 为其 System/360 选择了这种方法的变体）。1958 年夏天，他想出了一种新方法——用锗或硅制造所有单个组件，而不仅仅是晶体管。这与电子行业的主流经济趋势背道而驰，当时电阻器的价格很低，利润来自从生产成本中削减几十分之一美分。用硅制成的电阻器的成本要比用碳制成的电阻器高得多。但 Kilby 推断，如果电阻器和其他组件由与晶体管相同的材料制成，那么整个电路就可以用一块半导体材料制作而成。无论单个组件的成本增加了多少，都可以通过不再为每个组件设置单独的生产、包装和接线流程来抵消。

As others at TI and elsewhere were doing in 1958, Kilby looked at microminiaturization and made an assessment of the various government-funded projects then underway. Among those projects was one that TI was already involved with, called Micro-Module, which involved depositing components on a ceramic wafer.²⁶ Kilby did not find this approach cost effective (although IBM chose a variation of it for its System/360). In the summer of 1958 he came up with a fresh approach—to make all the individual components, not just the transistors, out of germanium or silicon. That swam against the tide of prevailing economics in the electronics business, where resistors sold for pennies, and profits came from shaving a few tenths of a cent from their production cost. A resistor made of silicon had to cost a lot more than one made of carbon. But Kilby reasoned that if resistors and other components were made of the same material as the transistors, an entire circuit could be fashioned out of a single block of semiconductor material. Whatever increased costs there were for the individual components would be more than offset by not having to set up separate production, packaging, and wiring processes for each.

1958 年 8 月，杰克·基尔比 (Jack Kilby) 制作了一个普通电路，电路中的所有元件（包括电阻和电容）都是用硅而不是通常的材料制作的。9 月，他制作了另一个电路，只不过这一次所有元件都由一块材料制成——一块 1/16 英寸 × 7/16 英寸的薄锗晶片。（该公司在这次演示中使用硅的能力还不足以胜任这项任务。）他和两名技术人员费力地在晶片上布置和构建了几个元件，并用细金线将它们相互连接起来。结果，振荡器工作了。1959 年初，他申请了专利，并于 1964 年获得批准（图 6.2）。27^德州仪器将其命名为“固体电路”。这是一项真正的创新，与当时军方资助的微型模块、分子电子学和其他小型化方案截然不同^。28

Jack Kilby built an ordinary circuit with all components, including its resistors and capacitor, made of silicon instead of the usual materials, in August, 1958. In September he built another circuit, only this time all the components were made from a single piece of material—a thin 1/16-inch × 7/16-inch wafer of germanium. (The company’s abilities to work with silicon for this demonstration were not quite up to the task.) He and two technicians laboriously laid out and constructed the few components on the wafer and connected them to one another by fine gold wires. The result, an oscillator, worked. In early 1959 he applied for a patent, which was granted in 1964 (figure 6.2).²⁷ Texas Instruments christened it the “solid circuit.” It was a genuine innovation, a radical departure from the military-sponsored micromodule, molecular electronics, and other miniaturization schemes then being pursued.²⁸

罗伯特·诺伊斯也是在中西部的爱荷华州格林内尔长大，他的父亲是那里的公理会牧师。有些人将诺伊斯的创造力归功于新教价值观，即持不同政见和寻找自己的救赎之路，²⁹但并非所有新教信仰都认同这一点，人们不会将诺伊斯或其他中西部发明家描述为宗教人士。更可能的解释是中西部农业社区特有的自给自足文化，尽管这个群体中只有一两位发明家真正在农场长大。无论如何，20 世纪 30 年代和 40 年代的玉米带是数字电子产品的沃土。

Robert Noyce also grew up in the Midwest, in Grinell, Iowa, where his father was a Congregational minister. Some ascribe Noyce’s inventiveness to Protestant values of dissent and finding one’s own road to salvation,²⁹ but not all Protestant faiths shared that, and one would not describe Noyce or the other Midwestern inventors as religious. A more likely explanation is the culture of self-sufficiency characteristic of Midwestern farming communities, even though only one or two of the inventors in this group actually grew up on farms. In any event, the Corn Belt in the 1930s and 1940s was fertile ground for digital electronics.

图 6.2

芯片。 ( a ) Jack Kilby 的集成电路专利。 ( b ) 平面晶体管。 (来源：飞兆半导体公司。) ( c ) Robert Noyce 的集成电路专利。

Figure 6.2

The chip. (a) Patent for integrated circuit by Jack Kilby. (b) Planar transistor. (Source : Fairchild Semiconductor.) (c) Patent for integrated circuit by Robert Noyce.

罗伯特·诺伊斯 (Robert Noyce) 正在加州山景城的仙童半导体公司工作，这时他听说了基尔比的发明。他一直在思考这个问题，并于 1959 年 1 月在实验室笔记本中描述了一个方案，该方案基本上与基尔比所做的相同，只是使用一块硅。30^他在仙童半导体公司的一位同事、瑞士出生的让·霍尔尼 (Jean Hoerni) 开辟了道路，他开发了一种制造硅晶体管的工艺，该工艺非常适合光刻生产技术，从而可以低成本地批量生产集成电路。31^它被称为“平面工艺”，顾名思义，它生产的晶体管是扁平的。（其他技术需要凸起的金属线或甚至以某种方式连接到表面的电线来连接晶体管。）该工艺最适合硅，因为可以构建氧化硅层——“人类已知的最佳绝缘体之一”，诺伊斯回忆道——并用于将一个设备与另一个设备隔离。³²对于诺伊斯来说，集成电路的发明并非是突然灵光一现的结果，而是他逐渐积累了有关材料、制造和电路的工程知识，这些知识大部分都是自 1957 年仙童公司成立以来积累的。（巧合的是，创立仙童半导体公司的资金来自一家相机公司，仙童相机和仪器公司。该公司以谢尔曼·费尔柴尔德 (Sherman Fairchild) 的名字命名，他是 IBM 最大的个人股东——他的父亲在本世纪初帮助创立了 IBM。）³³

Robert Noyce was working at Fairchild Semiconductor in Mountain View, California, when he heard of Kilby’s invention. He had been thinking along the same lines, and in January 1959 he described in his lab notebook a scheme for doing essentially the same thing Kilby had done, only with a piece of silicon.³⁰ One of his coworkers at Fairchild, Swiss-born Jean Hoerni, had paved the way by developing a process for making silicon transistors that was well-suited for photo-etching production techniques, making it possible to mass-produce ICs cheaply.³¹ It was called the “planar process,” and as the name implies, it produced transistors that were flat. (Other techniques required raised metal lines or even wires somehow attached to the surface to connect a transistor.) The process was best suited to silicon, where layers of silicon oxide—”one of the best insulators known to man,” Noyce recalled— could be built up and used to isolate one device from another.³² For Noyce the invention of the IC was less the result of a sudden flash of insight as of a gradual build-up of engineering knowledge about materials, fabrication, and circuits, most of which had occurred at Fairchild since the company’s founding in 1957. (By coincidence, the money used to start Fairchild Semiconductor came from a camera company, Fairchild Camera and Instrument. Sherman Fairchild, after whom the company was named, was the largest individual stockholder in IBM—his father helped set up IBM in the early part of the century.)³³

诺伊斯也在 1959 年 7 月申请了专利，比基尔比晚了几个月。多年后，法院最终解决了谁是“真正的”发明者这一争议，并给每个人及其公司分配了一定份额的荣誉。但大多数人都承认，诺伊斯的理念是将霍尔尼的平面工艺结合起来，这种工艺允许人们在制造设备本身的同时进行电气连接，这是后来集成电子学取得巨大进步的关键。

Noyce applied for a patent, too, in July 1959, a few months after Kilby. Years later the courts would sort out the dispute over who the “real” inventor was, giving each person and his respective company a share of the credit. But most acknowledge that Noyce’s idea to incorporate Hoerni’s planar process, which allowed one to make the electrical connections in the same process as making the devices themselves, was the key to the dramatic progress in integrated electronics that followed.

霍尔尼没有参与集成电路的专利，但他的贡献是众所周知的。“我可以走进世界上任何一家半导体工厂，看到我开发的东西被使用。这非常令人满意。” ³⁴他和诺伊斯的贡献说明了发明家如何首先通过视觉来解决问题，历史学家尤金·弗格森称之为“心灵之眼”。35^虽然这项发明需要对硅的物理和化学以及添加到其中的微量其他材料有透彻的了解，但它背后有一个非语言的视觉过程^。36

Hoerni did not share in the patents for the integrated circuit, but his contribution is well known. “I can go into any semiconductor factory in the world and see something I developed being used. That’s very satisfying.”³⁴ His and Noyce’s contributions illustrate how inventors cultivate a solution to a problem first of all visually, in what historian Eugene Ferguson calls the “mind’s eye.”³⁵ Although the invention required a thorough knowledge of the physics and chemistry of silicon and the minute quantities of other materials added to it, a nonverbal, visual process lay behind it.³⁶

这些 IC 发明步骤与空军或军方支持无关。飞兆半导体和德州仪器都不是首批获得空军小型化合同的公司。德州仪器早在锗被军事采用之前就率先将材料从锗转移到硅。基尔比使用单一材料制造传统设备的理念与空军的分子电子学和陆军的微模块概念相悖。平面工艺是飞兆半导体的内部创新^。37

These steps toward the IC’s invention had nothing to do with Air Force or military support. Neither Fairchild nor Texas Instruments were among the first companies awarded Air Force contracts for miniaturization. The shift from germanium to silicon was pioneered at Texas Instruments well before it was adopted for military work. Kilby’s insight of using a single piece of material to build traditional devices went against the Air Force’s molecular electronics and the Army’s micro-module concepts. And the planar process was an internal Fairchild innovation.³⁷

但集成电路发明后，美国航空航天界通过提供市场发挥了至关重要的作用。“先进”的民兵导弹是一种包裹在现有机身上的全新导弹。负责制导系统合同的北美航空公司分部 Autonetics 选择集成电路作为满足其要求的最佳方式。他们为其设计的计算机使用了大约 2,000 个集成电路和 4,000 个分立电路，而制导要求更简单的民兵一号则使用了 15,000 个分立电路。38 Autonetics^发表了两种电路的比较结果，以帮助支持他们的决定。据 Kilby 说，“在 20 世纪 60 年代初期，这些比较似乎非常引人注目，并且可能比其他任何事情都更有助于确立集成电路对军方的可接受性。” ³⁹民兵二号于 1964 年 9 月首次飞行；一年后，行业媒体报道称“民兵是最大的半导体用户”，每周生产六到七枚导弹。40^由于晶体管工厂产能过剩，该行业经历了繁荣与萧条的周期。如果没有民兵二号，他们就不会建立集成电路的批量生产线：“民兵的计划要求每周从德州仪器、西屋电气和 RCA 获得超过 4,000 个电路。” ⁴¹

But once the IC was invented, the U.S. aerospace community played a crucial role by providing a market. The “advanced” Minuteman was a brand-new missile wrapped around an existing airframe. Autonetics, the division of North American Aviation that had the contract for the guidance system, chose integrated circuits as the best way to meet its requirements. The computer they designed for it used about 2,000 integrated and 4,000 discrete circuits, compared to the 15,000 discrete circuits used in Minuteman I, which had a simpler guidance requirement.³⁸ Autonetics published comparisons of the two types of circuits to help bolster their decision. According to Kilby, “In the early 1960s these comparisons seemed very dramatic, and probably did more than anything else to establish the acceptability of integrated circuits to the military.”³⁹ Minuteman II first flew in September 1964; a year later the trade press reported that “Minuteman is top Semiconductor User,” with a production rate of six to seven missiles a week.⁴⁰ The industry had a history of boom and bust cycles caused by overcapacity in its transistor plants. Were it not for Minuteman II they would not have established volume production lines for ICs: “Minuteman’s schedule called for over 4,000 circuits a week from Texas Instruments, Westinghouse, and RCA.”⁴¹

费尔柴尔德公司并不是民兵的三大供应商之一。诺伊斯认为，军事合同扼杀了创新——他以空军的分子电子学为例，说明创新的方向是错误的。他尤其对这样一种看法感到不安：在军事资助下，

Fairchild was not among the three major suppliers for Minuteman. Noyce believed that military contracts stifled innovation—he cited the Air Force’s molecular electronics as an example of approaching innovation from the wrong direction. He was especially bothered by the perception that with military funding,

研究方向是由那些不太了解其发展方向的人决定的，研究人员自己花费了大量时间通过进度报告或访问等方式与军方人员进行沟通^。42

the direction of the research was being determined by people less competent in seeing where it ought to go, and a lot of time of the researchers themselves was spent communicating with military people through progress reports or visits or whatever.⁴²

然而不久之后，该公司就认识到了军事市场的价值：“军事和航天应用基本上占据了去年 [1963 年] 的整个集成电路市场，并将使用今年生产的电路的 95% 以上。” ⁴³

However, before long, the company recognized the value of a military market: “Military and space applications accounted for essentially the entire integrated circuits market last year [1963], and will use over 95 percent of the circuits produced this year.”⁴³

尽管不愿卷入军事合同，费尔柴尔德公司还是寻求机会向美国宇航局出售集成电路，用于其阿波罗制导计算机（图 6.3）。44^阿波罗计划的目标是在 20 世纪 60 年代末将人类送上月球，但这并非一项军事计划。其制导系统是麻省理工学院仪器实验室的产品，该实验室在查尔斯·斯塔克·德雷珀的领导下，还负责设计北极星和海神导弹的制导计算机。与民兵一样，阿波罗的设计人员一开始也只设计了适度的机载制导要求。最初大多数制导都将从地面处理；直到 1964 年，才开始使用模拟计算机。45^然而，随着登月任务的规模日益扩大，计算机经过重新设计，并被要求完成更多任务。该实验室是 1959 年第一批从 TI 购买集成电路的实验室之一。1961 年 8 月，NASA 选择仪器实验室负责阿波罗导航系统后，实验室的 Eldon Hall 与 TI 和 Fairchild 展开了谈判（图 6.4）。IC 体积小、重量轻，很有吸引力，尽管 Hall 担心缺乏可靠数量的生产数据。事后看来，这是一个不可避免的决定，他决定在计算机中使用 IC，采用 Fairchild 的“微逻辑”设计，使用 Philco-Ford、德州仪器和 Fairchild 生产的芯片。诺伊斯之所以选择费尔柴尔德的设计，可能是因为诺伊斯对麻省理工学院的代表很感兴趣。1961 年和 1962 年，麻省理工学院的代表曾多次拜访诺伊斯。（诺伊斯毕业于麻省理工学院。）⁴⁶美国宇航局于 1962 年 11 月批准了霍尔的决定，他的团队完成了一架原型机，该机于 1965 年 2 月首次运行，大约在民兵 II 首次飞行一年后^。47

Although reluctant to get involved in military contracts, Fairchild did pursue an opportunity to sell integrated circuits to NASA for its Apollo Guidance Computer (figure 6.3).⁴⁴ Apollo, whose goal was to put a man on the Moon by the end of the 1960s, was not a military program. Its guidance system was the product of the MIT Instrumentation Laboratory, which under the leadership of Charles Stark Draper was also responsible for the design of guidance computers for the Polaris and Poseidon missiles. Like Minuteman, Apollo’s designers started out with modest on-board guidance requirements. Initially most guidance was to be handled from the ground; as late as 1964 it was to use an analog computer.⁴⁵ However, as the magnitude of the Lunar mission manifested itself the computer was redesigned and asked to do a lot more. The lab had been among the first to purchase integrated circuits from TI in 1959. After NASA selected the Instrumentation Lab to be responsible for the Apollo guidance system in August 1961, Eldon Hall of the lab opened discussions with TI and Fairchild (figure 6.4). The IC’s small size and weight were attractive, although Hall was concerned about the lack of data on manufacturing reliable numbers of them in quantity. In a decision that looks inevitable with hindsight, he decided to use ICs in the computer, adopting Fairchild’s “micrologic” design with production chips from Philco-Ford, Texas Instruments, and Fairchild. His selection of Fairchild’s design may have been due to Noyce’s personal interest in the MIT representatives who visited him several times in 1961 and 1962. (Noyce was a graduate of MIT.)⁴⁶ NASA approved Hall’s decision in November 1962, and his team completed a prototype that first operated in February 1965, about a year after the Minuteman II was first flown.⁴⁷

与使用二十多种 IC 的民兵计算机相比，阿波罗计算机仅使用一种类型的 IC，采用简单的逻辑。48^每台阿波罗制导计算机包含大约 5,000 个这样的芯片。49^因此，微电子领域的当前“革命”很大程度上归功于民兵和阿波罗计划。民兵是第一个：它在集成电路发明几年后就将其用于关键应用。阿波罗迈出了下一个同样重要的步骤：它从一开始就被设计为利用集成逻辑的优势。

In contrast to the Minuteman computer, which used over twenty types of ICs, the Apollo computer used only one type, employing simple logic.⁴⁸ Each Apollo Guidance Computer contained about 5,000 of these chips.⁴⁹ The current “revolution” in microelectronics thus owes a lot to both the Minuteman and the Apollo programs. The Minuteman was first: it used integrated circuits in a critical application only a few years after they were invented. Apollo took the next and equally critical step: it was designed from the start to exploit the advantages of integrated logic.

阿波罗导航计算机共制造了约 75 台，其中约 25 台真正进入了太空。在此期间，从最初购买原型芯片到将其安装在阿波罗计算机的生产模型中，价格从每块 1,000 美元降至 20 至 30 美元之间。50^阿波罗合同，就像早先的民兵合同一样，为半导体公司提供了集成电路市场，而他们现在可以将这些集成电路卖给民用市场。到 1975 年最后一次阿波罗飞行（阿波罗-联盟号任务）时，一名宇航员携带了一台袖珍计算器（HP-65），其功能比机载计算机更强大。这就是航空航天界推动的创新步伐。

Around 75 Apollo Guidance Computers were built, of which about 25 actually flew in space. During that time, from the initial purchase of prototype chips to their installation in production models of the Apollo computer, the price dropped from $1,000 a chip to between $20 and $30.⁵⁰ The Apollo contract, like the earlier one for Minuteman, gave semiconductor companies a market for integrated circuits, which in turn they could now sell to a civilian market. By the time of the last Apollo flight in 1975 (the Apollo-Soyuz mission), one astronaut carried a pocket calculator (an HP-65) whose capabilities were greater than the on-board computer’s. Such was the pace of innovation set in motion by the aerospace community.

图 6.3

1969 年 7 月土星五号/阿波罗 11 号太空飞船发射。美国太空计划与计算机技术进步之间的关系非常复杂。阿波罗和民兵等计划的需求推动了微电子和计算机电路的发展。另一方面，计算技术的进步塑造了阿波罗等计划的设计和运行方式。（来源： NASA。）

Figure 6.3

Launch of the Saturn V/Apollo 11 spacecraft, July 1969. The relationship between the U.S. space program and the advance of computing technology was a complex one. The demands of programs like Apollo and Minuteman advanced the state of the art of microelectronics and computer circuits. Advances in computing, on the other hand, shaped the way programs like Apollo were designed and operated. (Source : NASA.)

图 6.4

麻省理工学院仪器实验室阿波罗计算机部门负责人埃尔登·霍尔 (Eldon Hall)，约 1968 年。（来源：查尔斯·斯塔克·德雷珀实验室。）

Figure 6.4

Eldon Hall, head of the Apollo Computer Division at the MIT Instrumentation Laboratory, ca. 1968. (Source : Charles Stark Draper Laboratory.)

芯片的商业影响

Commercial Impact of the Chip

这项发明对商业计算的最大影响是在小型计算机行业，而不是大型机行业。对于其 System/360 系列，IBM 开发了“固态逻辑技术”，这是一种类似于微模块的方案，其中电路沉积在约半英寸厚的陶瓷基板上，上面印有金属导电通道。51^到1964 年 4 月 360 发布时，集成电路正在迅速证明自己，IBM 的一些人担心它会被过时的技术所淘汰。1963 年 9 月撰写的一份 IBM 内部备忘录指出，IC“现在或未来五年都不会构成竞争威胁”，而 1964 年 9 月撰写的另一份内部报告认为，“单片机”（IBM 对 IC 的术语）已经取得了快速进展，IBM 在“实际经验上落后 2-4 年”，需要“6 个月到一年的时间才能赶上”IC 专业知识^。52

The biggest impact of this invention on commercial computing was in the minicomputer, not the mainframe, industry. For its System/360 line, IBM developed “solid logic technology,” a scheme similar to micro-module, in which circuits were deposited on a ceramic substrate about half an inch thick, with metallic conducting channels printed on it.⁵¹ By the time of the 360’s announcement in April 1964 the integrated circuit was rapidly proving itself, and some in IBM worried that it would be left behind with obsolete technology. An internal IBM memorandum written in September 1963 stated that ICs “do not constitute a competitive threat either now or in the next five years,” while another internal report written in September 1964 argued that rapid progress in “monolithics” (IBM’s term for ICs) had been made, and that IBM had a “2–4 year lag in practical experience” and needed “6 months to a year to catch up” in IC expertise.⁵²

两份备忘录都是正确的：IBM 已经学会了可靠地大批量生产固态逻辑技术电路，这在 System/360 中发挥了很好的作用。但为了保持竞争力，它在 1960 年代末发布的 System/370 中采用了 IC。但早在 1965 年春季首次交付 System/360 时，科学数据系统 (SDS) 就宣布了一款使用 IC 的计算机。当 RCA 决定用一系列兼容 360 的机器与 IBM 竞争时，它也决定采用集成电路。到 1966 年，SDS 和 RCA 计算机都开始交付，IBM 的“五年”领先优势现在变成了一年。集成电路从 Kilby 的粗糙实验室设备发展到商业计算机的实际应用，速度比预期的要快。部分原因是军事和航空航天客户的热切期望；这也要归功于 Noyce 的观察，即 IC 制造的基本技术是平面晶体管生产的革命性进步。

Both memorandums were right: IBM had learned to produce solid logic technology circuits reliably and in large quantities, which had served the System/360 well. But to remain competitive, it adopted ICs for the System/370, announced at the end of the decade. But as early as the first deliveries of the System/360 in the spring of 1965, Scientific Data Systems (SDS) had announced a computer that used ICs. When RCA decided to compete against IBM with a family of 360-compatible machines, it also decided to go with integrated circuits. By 1966, both SDS and RCA computers were being delivered and IBM’s “five year” lead was now one year. Integrated circuits went from Kilby’s crude laboratory device to practical use in commercial computers faster than anticipated. Part of the reason was the eagerness of military and aerospace customers; credit is also due to Noyce’s observation that the basic techniques of IC fabrication were an evolutionary advance over planar transistor production.

第二代小型计算机数字设备公司 (DEC) 展示了如何以适中的资本和适中的物理工厂进入计算机行业。随着价格低廉但功能强大的 IC 进入市场，其他公司可以效仿 DEC。DEC 并没有像 IBM 主宰大型机那样在小型计算机领域占据主导地位。与 BUNCH 不同，DEC 的竞争对手并不觉得他们必须回应每个产品公告，或提供软件兼容的产品。低成本、小巧的技术创新更为重要。结果是，从 1965 年到 1970 年代，小型计算机的性能以惊人的速度提升。价格保持在低位甚至下降。要进入这个市场，需要掌握最新技术，但银行和风险投资公司并不担心 — — 就像他们担心那些想与 IBM 竞争的人一样 — — Digital 会凭借其主导的市场份额击败新来者。

Second-Generation Minicomputers Digital Equipment Corporation showed how one could enter the computer business with a modest amount of capital and a modest physical plant. With inexpensive but powerful ICs on the market, the road opened up for others to follow DEC’s example. DEC did not dominate in minicomputers in the same way IBM dominated mainframes. Unlike the BUNCH, DEC’s competitors did not feel they had to answer every product announcement, or offer software-compatible products. Technical innovation, at low cost and in a compact package, mattered more. The result was that the performance of minis increased at a phenomenal rate from 1965 through the 1970s. Prices remained low and even dropped. To enter this market, one needed to have a grasp of the latest technology, but banks and venture capital firms did not fear—as they did with those who wanted to compete with IBM—that Digital would crush the newcomer by virtue of its dominant market share.

1968 年至 1972 年间，大约有 100 家新公司或老牌公司的分部在商业市场上推出小型计算机，五年间平均每三周推出一台。表 6.1列出了其中一些比较知名的公司^。53

Between 1968 and 1972, around one hundred new companies or divisions of established companies offered minicomputers on the commercial market, an average of one every three weeks for that five-year period. Table 6.1 lists some of the more well-known among them.⁵³

进入这个行业需要克服一些障碍，但大多数障碍都很低。54^半导体公司提供现成的廉价芯片，这些芯片在简单的封装中提供基本逻辑。到 1970 年，IC 制造商已经采用了一项标准，为计算机行业在未来二十年的发展奠定了坚实的基础。那就是提供一组使用晶体管实现逻辑的芯片，称为“晶体管-晶体管逻辑”(TTL)。TTL 芯片价格低廉，易于设计。55^它们封装在黑色塑料或陶瓷外壳中，引线沿着两侧排列，就像一条扁平的毛毛虫。仙童公司的工程师于 1964 年推出了这种“双列直插式封装”(DIP)。雷克斯·赖斯领导了这项工作；他的工作部分基于内森·普里蒂金 (Nathan Pritikin) 的专利工作（一位白手起家的发明家，后来因倡导低胆固醇饮食而闻名）。这种封装坚固耐用，易于操作^。56

There were a few barriers that one had to surmount to enter the business, but most barriers were low.⁵⁴ Semiconductor firms were offering a ready supply of inexpensive chips, which provided basic logic in a simple package. By 1970 the IC makers had adopted a standard that laid a solid foundation on which the computer industry would grow for the next two decades. That was to supply a set of chips that used transistors to implement the logic, called “transistor-transistor logic” (TTL). TTL chips were inexpensive and easy to design with.⁵⁵ They were packaged in a black plastic or ceramic case, with leads arranged along either side like a flattened caterpillar. Engineers at Fairchild introduced this “dual in-line package” (DIP) in 1964. Rex Rice led this effort; his work was based in part on work patented by Nathan Pritikin (a self-made inventor who later became famous for advocating a low-cholesterol diet). The package was rugged and easy to handle.⁵⁶

表 6.1

小型计算机，1965 - 1974 年

Table 6.1

Minicomputers, 1965–1974

制造商	电脑	年
加州数据处理器	十一/35	1974
辛辛那提米拉克龙	中压/2200	1970
计算机自动化	LSI“Naked Mini”系列	1972
电脑终端公司	数据点 2200
数据通用	新星	1969
	超新星	1971
数字计算机控制	DCC-116, 112	1972
数字设备公司	PDP-11 系列	1970
通用自动化	SPC-16	1971
通用电气	GEPAC 4010, 4020
GRI 计算机公司	GRI-99 系列	1972
GTE 信息系统	是/1000	1970
惠普	2100 系列	1971
霍尼韦尔	H-316	约 1970 年
	DDP-516	约 1971 年
IBM	系统 3	1969
数据互通	70 型
洛克希德电子	MAC-16	1968
	起诉	1972
模块化计算机系统	MODCOMP 线	1971
摩托罗拉	MDP-1000	1968
总理电脑	300 系列	1973
雷神公司	500 系列	1974
科学数据系统	SDS-910	1962
	SDS-920	1965
系统工程实验室	SEL-810
德州仪器	960, 980	1974
瓦里安联合公司	520
	620	1972
西屋电气	W-2500	1971

到 1970 年，芯片之间的连接方式也已标准化。由 Globe-Union 率先发明的印刷电路板也已发展到可以处理集成电路。小型计算机设计师现在可以设计一块大型印刷电路板，上面有小型计算机电路所需的所有 IC 的位置。在一条装配线上（可能位于亚洲，由女性操作以节省劳动力成本），一个人（或机器）会将芯片“塞入”电路板一侧的孔中。然后，她将电路板放入一个腔室中，一股熔化的焊料会滑过突出到另一侧的引脚，将它们牢固地连接到印刷连接上。该过程快速、可靠，并且能生产出坚固耐用的产品。仙童公司的工程师在 20 世纪 60 年代末首次在名为 SYMBOL 的实验计算机上探索了这种封装和布线方法。SYMBOL 的建立是为了探索一种新的计算机架构；事实证明，它的封装和布线比其架构更具影响力。⁵⁷这种组合——TTL 逻辑、双列直插式封装和大型波峰焊印刷电路板——在 20 世纪 90 年代仍然是标准^。58

By 1970 a way of connecting the chips to one another had also standardized. The printed circuit board, pioneered by Globe-Union, had evolved to handle integrated circuits as well. A minicomputer designer could now lay out a single large printed circuit board, with places for all the ICs necessary for the circuits of a small computer. On an assembly line (possibly located in Asia and staffed by women to save labor costs), a person (or machine) would “stuff” chips into holes on one side of the board. She would then place the board in a chamber, where a wave of molten solder would slide across the pins protruding through to the other side, attaching them securely to the printed connections. The process was fast, reliable, and yielded a rugged product. It was Fairchild engineers who first explored this method of packaging and wiring, on an experimental computer called SYMBOL, in the late 1960s. SYMBOL was built to explore a new computer architecture; its packaging and wiring proved far more influential than its architecture.⁵⁷ This combination—TTL logic, dual in-line packaging, and large, wave-soldered printed circuit boards—remained a standard into the 1990s.⁵⁸

英特尔数字设备公司的成立以其直接内存访问等架构上的进步而闻名。第二波小型计算机产品展示了同样引人注目的架构创新。第一个是将小型计算机的字长定为 16 位。这一举措紧随 IBM 的脚步，IBM 于 1964 年宣布推出 System/360。System/360 使用 32 位字，但它也设定了 8 位作为“字节”的标准，即对字母或其他字符进行编码的基本位组。8 位字节允许 256 种不同的组合 - 远远超过数字、标点符号和大小写字母所需的组合 - 但 8 是 2 的幂，这简化了计算机设计的某些方面。而“额外”的组合为未来的发展留下了空间。⁵⁹ IBM 选择的 8 位字节成为了标准，以至于很少有人记得 6 位曾经是小型计算机制造商的标准，对于他们来说，“额外”的位占了整个系统成本的很大一部分。

The Founding of Intel Digital Equipment Corporation made its mark by architectural advances like Direct Memory Access. The second wave of minicomputer products showed an equally-remarkable set of architectural innovations. The first was to settle on sixteen bits for the mini’s word length. This move followed the lead of IBM, after its announcement of the System/360 in 1964. System/360 used a 32-bit word, but it also set a standard of 8 bits as the “byte,” the basic group of bits to encode a letter of the alphabet or other character. An 8-bit byte allowed 256 different combinations—far more than needed for numbers, punctuation and the upper and lower case alphabet—but eight was a power of two, which simplified certain aspects of a computer’s design. And the “extra” combinations left room for future growth.⁵⁹ IBM’s choice of an 8-bit byte became a standard, so much so that few remember that 6 bits had been the standard among minicomputer manufacturers, for whom the “extra” bits made up a proportionally greater cost of the overall system.

有一个区别：小型计算机不使用 IBM 的 EBCDIC 代码来表示字符，而是开发了 8 位 ASCII 扩展。这一决定产生了一些长期影响。它扩大了 IBM 与其他计算机行业之间的差距。由于 ASCII 的第 8 位没有标准化，因此也导致小型计算机以及后来的个人计算机的不同标准代码激增。这种分裂的最终后果将在 20 世纪 80 年代显现出来，当时 IBM 带着 ASCII 机器进入了个人计算机市场。

There was one difference: minicomputers did not use IBM’s EBCDIC code for a character but developed an 8-bit extension of ASCII instead. This decision had some long-term consequences. It widened the gap between IBM and the rest of the computer industry. Because the eighth bit of ASCII was not standardized, it led also to a proliferation of different standard codes for mini, and later personal, computers. The ultimate consequences of this split would play out in the 1980s, after IBM entered the personal computer market with an ASCII machine.

DEC 的 PDP-8 具有 12 位字长，每个字符有 6 位代码，是在 360 发布后推出的。对于竞争对手 Computer Controls Corporation 的 14 位小型机工程师 Gardner Hendrie 来说，IBM 宣布推出 System/360 是一个重磅消息。1964 年夏天，CCC 重新设计了它的机器，并于 10 月发布了第一台 16 位小型机 DDP-116。60^这台机器为 CCC 带来了利润。该公司发展迅速，一度看起来会超越数字设备公司，但 1966 年霍尼韦尔收购了 CCC，CCC 失去了独立性。霍尼韦尔继续生产 16 位产品线，但也许是因为它将更多的注意力放在了大型机上，产品线逐渐萎缩。

DEC’s PDP-8, with its 12-bit word length and a 6-bit code for each character, was shipped after the 360’s announcement. For Gardner Hendrie, an engineer working on a 14-bit mini at the rival Computer Controls Corporation, IBM’s announcement of the System/360 was a bombshell. Over the summer of 1964 CCC redesigned its machine, and in October it announced the DDP-116, the first 16-bit minicomputer.⁶⁰ The machine was profitable for CCC. The company grew rapidly and for a while it looked like it would overtake Digital Equipment Corporation, but in 1966 Honeywell bought CCC, and it lost its independence. Honeywell continued the 16-bit line, but perhaps because it was devoting more attention to mainframes, the product line withered.

霍尼韦尔进军小型计算机市场最终只是一次小插曲。如果不是因为与霍尼韦尔小型计算机有关的一次特殊事件，没有人会记得它。从 1967 年开始，美国国防部高级研究计划局 (ARPA) 开始召开一系列会议，讨论如何将全国各地的计算机连接到一个网络中。61 ^ARPA想要建立这样一个网络有几个原因，主要原因是希望在计算机体积庞大且价格昂贵的时代，在接受其资助的地方之间共享资源。在这个过程的早期，ARPA 的研究人员就意识到连接的计算机的多样性是一个主要问题。他们建议通过在每个节点（联网的主计算机和网络本身之间）使用一台小型计算机来解决这个问题。对于这个接口消息处理器 (IMP)，他们选择了霍尼韦尔 DDP-516，这是 16 位 CCC mini 的后代。62 ¹⁹⁶⁹年 12 月，四个节点（即四个 IMPS）在落基山脉以西运行；一年后，这个数字增加到了十个，遍布全国。到 1971 年^， ARPANET 已由连接 23 台主机的 15 个节点组成。其中 9 台是 PDP-10，5 台是 IBM System/360，1 台是 Illiac-IV，其余的是各种小型计算机和大型机。63 ARPANET 于 1972 年 10 月在华盛顿特区的一家酒店公开展示；到那一年，已经有 30 个节点。64 ^ARPANET于 1988 年解散，但它将永远被人们铭记为互联网的前身，互联网在 20 世纪 90 年代进入公众视野。到那时，IMPS 已不再需要连接节点；它们被一组软件和硬件开发所取代，使它们变得不再必要。几台 DDP-516 计算机被保存下来，并被精心保存在博物馆和私人收藏中，作为 19 世纪 40 年代塞缪尔·莫尔斯电报设备的现代版本^。65

Honeywell’s foray into the minicomputer market ended up as a minor diversion. No one would remember Honeywell minicomputers were it not for one exceptional event associated with them. Beginning in 1967 the Advanced Research Projects Agency (ARPA) of the Department of Defense began a series of meetings to discuss how to link computers across the country in a network.⁶¹ ARPA had several reasons for wanting such a network, the main one being a desire to share resources among places receiving its funding at a time when computers were large and expensive. Early in the process the ARPA researchers recognized that the diverse nature of the computers being connected was a major problem. They proposed to solve it by using a minicomputer at each node, between the main computer being networked and the network itself. For this interface message processor (IMP), they chose a Honeywell DDP-516, a descendent of the 16-bit CCC mini.⁶² Four nodes (i.e., four IMPS) were operating west of the Rockies in December 1969; a year later there were ten, spanning the country. By 1971 ARPANET consisted of fifteen nodes connecting twenty-three host computers. Of those, nine were PDP-10s, five were IBM System/360s, one was the Illiac-IV, and the rest were assorted minicomputers and mainframes.⁶³ ARPANET was demonstrated publicly at a hotel in Washington, D.C., in October 1972; by that year there were thirty nodes.⁶⁴ ARPANET was dismantled in 1988, but it will always be remembered as the precursor of the Internet, which burst into the public’s consciousness in the 1990s. By that time IMPS were no longer needed to connect the nodes; they were replaced by a set of software and hardware developments that made them unnecessary. A few DDP-516 computers were saved and have been carefully preserved in museums and private collections as the modern equivalent of Samuel Morse’s telegraph apparatus of the 1840s.⁶⁵

另一家很快意识到 16 位字长优势的公司是 Data General。Digital Equipment Corporation 的工程师 Edson DeCastro 对 DEC 的 16 位计算机计划感到失望，DEC 将其称为 PDP-X。开发 16 位机器是一项重大任务，因为 DEC 非常清楚 Computer Controls Corporation 的成功。DeCastro 提出了 PDP-X 的设计，但 DEC 的管理层拒绝了。1968 年春天，他和另外两名 DEC 工程师辞职，在邻近的哈德逊镇创立了 Data General。他们的目标是生产一台 16 位迷你电脑^。66

Another company that quickly recognized the advantages of a 16-bit word length was Data General. Edson DeCastro, an engineer at Digital Equipment Corporation, grew frustrated with DEC’s plans for a 16-bit computer, which DEC was calling the PDP-X. Developing a 16-bit machine was a major assignment, as DEC was well aware of Computer Controls Corporation’s success. DeCastro proposed a design for the PDP-X, but DEC’s management turned it down. In the spring of 1968 he and two other DEC engineers resigned and founded Data General in the neighboring town of Hudson. Their goal was to produce a 16-bit mini.⁶⁶

作为数字设备公司的后代，Data General 的成立反映了一种后来定义了计算机文化的现象，尤其是在西海岸的硅谷。一位观察家评论道：“在这个行业中，……传统意义上的厂房、设备和原材料等资本资产毫无意义……智力就是全部。” ⁶⁷ Ken Olsen 却不这么认为。他认为 DeCastro 在 DEC 工作期间开发了 PDP-X，现在正在使用该设计推出 Data General。然而，Data General 在 1968 年底发布的 16 位机器 Nova 并不是被拒绝的 PDP-X。它更简单、更优雅。那些使用过它的人仍然称它为“干净的机器”，这是真正优化设计的罕见例子——复杂，但不会比必要的更复杂^。68

Data General’s founding as a descendent of Digital Equipment echoed a phenomenon that came to define the computer culture, especially in Silicon Valley on the West Coast. One observer remarked: “In this business, .. . capital assets in the traditional sense of plant, equipment, and raw materials counted for nothing. ... Brainpower was the entire franchise.”⁶⁷ Ken Olsen felt otherwise. He believed that DeCastro had developed the PDP-X while a DEC employee and now was using that design to launch Data General. The Nova, the 16-bit machine that Data General announced at the end of 1968, was not the rejected PDP-X, however. It was simpler and more elegant. Those who used one still talk of it as a “clean machine,” a rare example of a truly optimal design—complex, but no more complex than necessary.⁶⁸

Nova 还采用了上一年没有的电路和封装方面的进步。它使用的芯片包含的晶体管比早期计算机使用的更多。后来，为了与前代产品区分开来，这些芯片被称为中规模集成 (MSI)。（后来出现了大规模集成 [LSI]，最后是超大规模集成 [VLSI]，基于芯片上的晶体管数量。）这些芯片依次安装在约 15 平方英寸的单个印刷电路板上。这块电路板比当时的标准电路板大，因此封装非常紧凑。

The Nova also incorporated advances in circuits and packaging not available the previous year. It used chips that contained more transistors than earlier computers had used. These were later called medium scale integration (MSI) to distinguish them from their predecessors. (Later on came large scale integration [LSI] and finally very large scale integration [VLSI], based on the number of transistors on a chip.) These, in turn, were mounted on a single printed circuit board about 15 inches square. This board was larger than what was standard at the time and made for a very compact package.

1971 年中，该公司推出了一款先进的“Super Nova”，它采用了另一项技术创新，即用集成电路代替磁芯作为随机存取存储器 (RAM)。尽管制造用于存储信息的 IC 一直是可能的，但由于 SAGE 积累了生产经验，磁芯存储器的价格大约便宜了十倍。1970 年，伊利诺伊大学的一个超级计算机项目选择了 Fairchild 的 256 位内存芯片作为中央内存（图 6.5），这是一个突破。Illiac-IV 是一个雄心勃勃的尝试，旨在突破冯·诺依曼顺序处理架构。Illiac-IV 遇到了许多困难，包括厄巴纳校园的学生骚乱，以及实施雄心勃勃的设计的问题。它最大的遗产可能是它使用了半导体存储器，这为 Data General 的商业用途铺平了道路^。69

In mid-1971 the company introduced an advanced “Super Nova,” which incorporated still another technical innovation, integrated circuits instead of magnetic cores for random access memory (RAM). Although it had always been possible to make ICs that would store information, core memories had been about ten times cheaper because of the experience in producing them, going back to SAGE. A breakthrough came in 1970, when a supercomputer project at the University of Illinois chose 256-bit memory chips from Fairchild for its central memory (figure 6.5). The Illiac-IV was an ambitious attempt to break out of the von Neumann architecture of sequential processing. The Illiac-IV was beset with many difficulties, including student unrest on the Urbana campus, as well as problems implementing the ambitious design. Its greatest legacy was probably its use of semiconductor memory, which paved the way for Data General’s commercial use.⁶⁹

Super Nova 确立了半导体 RAM 在商用计算机中的可行性。这只是硅革命即将发生的几个指标之一。我们已经提到过 IBM 决定为其 System/370 采用单片内存。在 Illiac-IV、System/370 或 Super Nova 出现之前，发生了一件更为重大的事件。1968 年 7 月，罗伯特·诺伊斯和戈登·摩尔离开仙童公司，成立了一家新公司，该公司最初的重点是半导体内存。再一次，“智力就是全部”：诺伊斯和摩尔是仙童半导体部门 1957 年的最初创始人之一，他们离开了威廉·肖克利创立的一家公司，该公司开始在旧金山半岛南部生产晶体管。这两个人，很快与仙童公司的安德鲁·格罗夫一起加入，他们选择了一个“集成电子”一词的缩写名称：英特尔。⁷⁰ 1970 年，英特尔推出了 1,024 位动态 RAM 芯片 1103。这款芯片以及 Super Novas 和 IBM System/370 的推出标志着磁芯时代的终结。71^英特尔从这个不起眼的起点开始，成为 20 世纪 80 年代和 90 年代计算领域的主导力量。（其后续发展将在后面的章节中讨论）。

The Super Nova established semiconductor RAM’s viability for commercial computers. It was only one of several indicators that a revolution in silicon was about to happen. We have already mentioned IBM’s decision to go to monolithic memory for its System/370. Before the Illiac-IV, System/370, or the Super Nova appeared, an event took place that would be even more significant. In July 1968 Robert Noyce and Gordon Moore left Fairchild and founded a new company, whose initial focus would be on semiconductor memory. Once again, “Brainpower was the entire franchise”: Noyce and Moore had been among the original founders of Fairchild’s semiconductor division in 1957, having left a company founded by William Shockley that began transistor production in the lower San Francisco peninsula. The two men, soon joined by Andrew Grove, also of Fairchild, chose a name that would be a contraction of the words “integrated electronics”: Intel.⁷⁰ In 1970 it introduced a 1,024-bit dynamic RAM chip, the 1103. That, plus deliveries of Super Novas and IBM System/370s, marked the beginning of the end for magnetic core.⁷¹ From that modest beginning, Intel would become a dominant force in computing in the 1980s and 1990s. (Its subsequent path will be discussed in following chapters).

（A）

(a)

（二）

(b)

（C）

(c)

图 6.5

( a ) 飞兆半导体存储器芯片，用于 Illiac-IV，是磁芯的第一个实用替代品。该芯片存储 256 位。（来源：飞兆半导体。）( b ) Data General Super Nova。（来源：Data General Corporation。）( c ) 英特尔 1103 存储器芯片。容量 1,024 位。（来源：查尔斯·巴贝奇研究所，

Figure 6.5

(a) Fairchild memory chip, used in Illiac-IV, and the first practical alternative to magnetic core. The chip stored 256 bits. (Source : Fairchild Semiconductor.) (b ) Data General Super Nova. (Source : Data General Corporation.) (c) Intel 1103 memory chip. Capacity 1,024 bits. (Source : Charles Babbage Institute,

Super Nova 被封装在一个紧凑的矩形盒子中。前面板上有一排指示灯，用于指示内部寄存器的状态，并排有开关，用于设置或重置这些寄存器中的单个位。后来的计算机取消了前面板。除了这一例外，Nova 的硬件配置（矩形盒子、大型印刷电路板以及用于处理器和内存的芯片）一直沿用至今。

The Super Nova was packaged in a compact rectangular box. On the front panel were rows of lights indicating the status of its internal registers, with parallel rows of switches that allowed one to set or reset individual bits in those registers. Later generations of computers would do away with the front panel. With that exception, the hardware configuration of the Nova—a rectangular box, large printed circuit boards, and chips for both the processor and memory—has persisted to the present day.

PDP-11到 1969 年，数字设备公司发现自己处于劣势。它曾推出过使用集成电路的 PDP-8 型号 (PDP 8/I)，但其 12 位字长和有限的指令库与 CCC/Honeywell 或 Data General 的机器不相上下。其他大大小小的公司也纷纷进入该领域。72^受到DeCastro 叛逃的震惊，DEC 再次尝试设计 16 位小型机。1969 年 3 月，四名 DEC 工程师飞往匹兹堡，与 Gordon Bell 和同样是该校教授的 William Wulff 进行磋商。Bell 已暂时离开 DEC，前往卡内基梅隆大学任教。卡内基教授们对提出的设计并不热衷。DEC 工程师 Harold McFarland 提出的替代设计更有希望。该小组“决定放弃大约一年的工作成果”，并根据麦克法兰的想法重新设计计算机。73^DEC将新机器称为 PDP-11，并于 1970 年 1 月发布，并在几个月后开始交付。

The PDP-11 Digital Equipment Corporation found itself at a disadvantage by 1969. It had introduced a model of the PDP-8 that used integrated circuits (the PDP 8/I), but its 12-bit word length and limited instruction repertoire did not compare well to the machines from CCC/ Honeywell or Data General. Other companies, large and small, were also entering the field.⁷² Stung by DeCastro’s defection, DEC made another try at designing a 16-bit mini. In March 1969 four DEC engineers flew to Pittsburgh to consult with Gordon Bell, who had taken a temporary leave from DEC to teach at Carnegie Mellon University, and with William Wulff, also a professor there. The Carnegie professors were not enthusiastic about the proposed design. An alternate design proposed by one of the DEC engineers, Harold McFarland, showed more promise. The group “decided to discard about a year’s worth of work” and redesign the computer around McFarland’s ideas.⁷³ DEC called the new machine the PDP-11, announced it in January 1970, and began deliveries a few months later.

在重新设计的过程中，团队想出了一个创新，使其重新夺回了被 Data General 抢走的市场。那就是对总线概念的重新定义，总线是一组以通用和标准方式为机器所有主要部分提供服务的电线。这个概念并不新鲜。20 世纪 40 年代安装在哈佛的机电 Mark I 就使用了一个总线（拼写为“buss”）。⁷⁴ Whirlwind 也有总线，Nova 和 PDP-8 型号也有。但 PDP-11 的 Unibus 将这个概念进一步发展：机器的几乎所有主要单元，包括内存和 I/O 设备，都连接到一个通用的 56 线总线。这使得 DEC 及其客户特别容易为专门的应用程序配置或扩展机器^。75

In the course of the redesign, the team came up with an innovation that allowed it to regain the ground it had lost to Data General. That was a redefinition of the concept of a bus, a set of wires that served all major sections of the machine in a common and standard way. The notion was not new. The electromechanical Mark I, installed at Harvard in the 1940s, used one (spelled “buss”).⁷⁴ The Whirlwind also had a bus, as did the Nova, and models of the PDP-8. But the PDP-11’s Unibus carried this concept further: nearly all major units of the machine, including memory and I/O devices, were connected to a common, 56-line bus. That made the machine especially easy for DEC as well as its customers to configure for specialized applications or to expand.⁷⁵

总线架构，就像 Nova 开创的封装一样，从那时起就在计算机设计中占据了主导地位。76^这一代的 16 位字长将翻倍为 32 位，然后是 64 位，但它仍然是 2 的幂。

The bus architecture, like the packaging pioneered by the Nova, has prevailed in computer design ever since.⁷⁶ The 16-bit wordlength of this generation would double to 32, and then 64 bits, but it would remain a power of two.

PDP-11 迅速超越了竞争对手，并继续推动着数字设备公司的发展。PDP-8 的销售使公司员工人数从 1965 年的 900 人增加到 1970 年的 5,800 人。在 PDP-11 的帮助下，到 1977 年，公司员工人数增长到 36,000 人。77 ²⁰世纪 70 年代，PDP-11 销量超过 170,000 台。78 ¹⁹⁷⁰年左右，计算机行业遭遇经济衰退，但 DEC 凭借其 PDP-11 和 Data General 凭借 Nova 幸存下来并蓬勃发展。竞争对手纷纷倒下或只找到了小众市场。繁荣时期已经结束（它们会卷土重来），但它们留下了对计算机行业及其技术的重新定义。

The PDP-11 quickly surpassed its rivals and continued to fuel Digital Equipment Corporation’s growth. Sales of the PDP-8 had enlarged the company from about 900 employees in 1965 to 5,800 in 1970. With the help of the PDP-11, it grew to 36,000 employees by 1977.⁷⁷ Over 170,000 PDP-11’s were sold in the 1970s.⁷⁸ A recession hit the computer industry around 1970, but DEC, with its PDP-11, and Data General, with the Nova, survived and prospered. Competitors fell by the wayside or found only small niche markets. The go-go years were over (they would return), but they left in their wake a redefinition of the computer industry and its technology.

Nova 的成功源于其优雅的设计和创新的包装；PDP-11 的成功源于其创新的架构，这种架构使小型计算机能够实现以前属于大型计算机领域的大量应用程序。除了 Unibus 之外，该机器还采用了许多寻址“模式”，允许人们以各种灵活而强大的方式访问数据。“Digital 的传统业务是销售架构，”当时在 DEC 工作的一位工程师说道。在他看来，PDP-11 的架构“很棒”，尽管该公司实现该架构的方式与 Data General 的做法相比过于复杂。79^它比 PDP-8 复杂得多。根据 Gordon Bell 的说法，“PDP-11 最初是一款难以理解的机器，只有那些拥有丰富计算机经验的人才能买到。” ⁸⁰

The Nova’s success came from its elegant design and innovative packaging; the PDP-11’s from its innovative architecture, which opened up minicomputers to a host of applications that had previously been the domain of mainframe computers. In addition to the Unibus, the machine employed a number of addressing “modes,” which allowed one access to data in a variety of flexible and powerful ways. “Digital’s traditional business is to sell architecture,” said one engineer who worked for DEC in those years. In his view, the PDP-11’s architecture was “wonderful,” although the way the company implemented it was overly complex compared to the way Data General did things.⁷⁹ It was a much more complex machine than the PDP-8. According to Gordon Bell, “The PDP-11 was initially a hard machine to understand and was marketable only to those with extensive computer experience.”⁸⁰

PDP-11 的强大功能意味着那些真正了解它的人可以开发软件，让其他人可以轻松使用，这为这一代小型计算机提供了另一种区分方式：用户使用熟悉的语言（如 FORTRAN）而不是机器代码进行编程，对于 PDP-8 用户来说，这种活动似乎近乎黑魔法。新的小型计算机还配备了允许轻松编辑和简化程序中错误查找和更正的工具。对于许多客户来说，这是两全其美的选择——既有分时系统的灵活性和易用性，又没有分时大型机的开销和费用。

The PDP-11’s power meant that those who did understand it could develop software to make it easy for others to use, which suggests another way to distinguish this generation of minicomputers: users programmed them in familiar languages like FORTRAN rather than in machine code, an activity that seemed close to black magic for users of the PDP-8. The new minicomputers also came with tools that allowed easy editing and simplified finding and correcting bugs in programs. For many customers it was the best of both worlds—the flexibility and ease of use of a time-sharing system, with none of the overhead and expense of time-sharing a mainframe.

直接访问计算的胜利

Direct-Access Computing Triumphant

在小型计算机系统日趋成熟的同时，大型机系统也在不断发展，以提供类似的交互式功能。在 20 世纪 60 年代后期，大型机的分时功能非常困难；到 20 世纪 70 年代中期，分时软件包已成为许多 System/370 安装中强大而稳定的组成部分。除了对话监控系统 (CMS) 之外，IBM 从 1971 年开始为其 370 计算机提供分时选项 (TSO)，而通用电气/霍尼韦尔则基于其在达特茅斯学院的工作（稍后讨论）提供了成功的商业系统。对于大型和小型系统而言，分时成为一种可接受且经济的计算机使用方式。

While minicomputer systems were maturing, mainframe systems were evolving to offer similar, interactive capabilities. Time-sharing a mainframe was difficult in the late 1960s; by the mid-1970s, time-sharing packages were robust and stable parts of many System/370 installations. In addition to its Conversational Monitoring System (CMS), IBM offered a time sharing option (TSO) for its 370 computers beginning in 1971, while General Electric/Honeywell offered a successful commercial system based on its work at Dartmouth College (discussed later). For large and small systems, time-sharing became an acceptable and economical way of using a computer.

一个关键因素是磁盘存储的发展，它提供了对大量数据的快速和直接访问。IBM 在 20 世纪 50 年代末率先使用 RAMAC 进行磁盘存储，但在接下来的十年里，顺序访问磁带（而不是磁盘）仍然是大型机大容量存储的主流。随着 System/370 的推出，IBM 推出了新的磁盘存储模型，性能得到了显著提高。在 20 世纪 60 年代末和 70 年代初，在磁盘上存储数据的成本下降了 20 倍，而典型磁盘存储系统的容量增加了 40 倍^。81

A key factor was the development of disk storage that offered rapid and direct access to large amounts of data. IBM had pioneered the use of disk storage with RAMAC in the late 1950s, but for the next ten years sequentially accessed tape, not disks, remained the mainstay of mass storage on mainframes. With the System/370, IBM introduced new models of disk storage that offered dramatically increased performance. During the late 1960s and early 1970s, the cost of storing data on disks dropped twentyfold, while the capacity of a typical disk storage system increased fortyfold.⁸¹

在许多装置中，磁带卷轴与磁盘组相连，它们看起来像蛋糕盒，作为大容量存储介质。到 1980 年，使用不可移动磁盘的驱动器变得很普遍，如果可以移动，则用读写头密封，以保持严格的公差。IBM 发明了一种驱动器，最初有两个主轴，每个主轴可容纳 30 兆字节的数据：人们称它们为“温彻斯特”驱动器，以 30-30 步枪命名。IBM 将磁盘称为“直接访问存储设备”，缩写为 DASD。这些产品的商业成功促使其他公司纷纷推出与 IBM 可插拔的磁盘驱动器，这再次引发了华尔街的一系列活动，并不可避免地引发了针对 IBM 的反垄断诉讼^。82

At many installations, the tape reels were joined by disk packs—they looked like cake boxes—as mass storage media. By 1980 it became common to use drives with disks that were not removable, or if they were, they were sealed with their read-write heads, to maintain tight tolerances. IBM came up with a drive that initially had two spindles holding 30 megabytes of data each: people called them “Winchester” drives, after the 30-30 rifle. IBM called disks “Direct Access Storage Devices,” shortened to the acronym DASD. The commercial success of these products led other companies to rush in with disk drives that were plug-compatible with IBM’s, leading once again to a flurry of activity on Wall Street and the inevitable lawsuits against IBM for alleged antitrust activity.⁸²

直接访问计算成功的一个鲜为人知的因素是 IBM 软件产品 CICS，该产品于 20 世纪 60 年代末为电力行业开发，但很快便在整个商业领域流行起来。83 ^CICS并非针对任何特定应用程序量身定制，而是允许客户编写自己的程序，以便直接查询和检索大型数据库中的数据。最早利用 CICS 的是燃气和电力公司，它们可以用它来通过电话回答客户的疑问。它也改变了美国的零售业。例如，考虑一下一年四季日夜发生的典型交易：某人打电话给缅因州的一家邮购公司，询问一双鞋的价格、颜色、尺码和是否有货，确认订单，用信用卡付款，并订购当天下午发货以便隔夜送达。通过提供印在目录上的客户编号，公司可以确定此人的正确地址（包括邮政编码）以及信用记录和近期购买模式。检查并扣取此人的信用卡账户；电话公司向邮购公司收取电话费，库存记录得到更新，也许还会向亚洲的一家工厂发送订单，以生产更多的鞋子。隔夜承运人会收到发货通知，并生成发票。整个交易只需几分钟。大多数库存和账单信息直接从一台计算机传输到另一台计算机。只有少量纸质记录。许多此类操作都使用 CICS 或类似软件。这种活动变得如此普遍，以至于市中心百货商店的传统零售采购越来越被视为例外。尽管来自小型计算机和工作站的竞争日益激烈，但大型机处理此类数据的能力使其保持了可行性。

A little-recognized factor in the triumph of direct access computing was the IBM software product CICS, developed in the late 1960s for the electric utility industry but soon found throughout the commercial world.⁸³ CICS was not tailored to any specific application but rather allowed customers to write their own programs to permit direct query and retrieval of data from large databases. Among the first to exploit it were gas and electric utilities, who could use it to answer customer queries over the telephone. It has also transformed retail sales in the United States. Consider, for example, a typical transaction that occurs day and night, year round: a person calls a mail order house in Maine, asks about price, color, size, and availability of a pair of shoes, confirms the order, pays by credit card, and orders it to be shipped that afternoon for overnight delivery. By giving out a customer number printed on the catalog, the company determines the person’s correct address including ZIP code, as well credit history and recent buying patterns. The person’s credit card account is checked and debited; the telephone company bills the mail-order house for the telephone call, the inventory record is updated, and perhaps an order is sent to a factory in Asia to make more shoes. The overnight carrier is notified of the shipment, and an invoice for that is generated as well. The whole transaction takes a few minutes. Most of the inventory and billing information is transferred directly from one computer to another. There are only a few paper records. CICS, or software similar to it, is used in many of these operations. This kind of activity has become so common that traditional retail buying at a downtown department store is increasingly seen as exceptional. The mainframe’s ability to handle this kind of data kept it viable in spite of the increasing competition from minicomputers and workstations.

计算机科学教育

Computer Science Education

受这些趋势影响的另一个地方是学术界，20 世纪 70 年代的技术进步彻底改变了计算机编程的研究和教学。

Another place affected by these trends was the academic world, where the technological advances of the 1970s transformed both research and the teaching of computer programming.

尽管自 20 世纪 50 年代以来，大学就一直在教授计算机课程，但批处理环境并不是最佳选择。当程序本身没有错误时，批量运行程序的效果最好。在 NASA-Ames 研究中心等计算机处理风洞数据的设施中，程序尽快处理每天的数据非常重要。在最初的调试期之后，程序中存在的所有错误都会被发现和纠正。编译器是一个复杂的程序，占用了大量的计算机内存和时间，但这些时间花得很值得，因为编译器生成的机器码执行速度很快。

Although universities had been teaching courses on computing since the 1950s, a batch environment was hardly optimum for that. Running batches of programs worked best when the programs themselves were error-free. In an installation like the NASA-Ames Research Center, where the computer-processed wind tunnel data, it was important that the program plow through each day’s data as fast as possible. After an initial shakedown period, whatever errors the programs had were found and corrected. The compiler was a sophisticated program that used a lot of the computer’s memory and time, but that time was well spent since the compiler generated machine code that executed quickly.

在教学环境中，情况正好相反：程序必然会包含错误。由于程序很短，通常不处理大量数据，因此编译后的程序执行速度并不重要。由于许多学生会提交不同的程序，因此编译器的运行速度很重要，并且除了机器代码外，它还会对发现的错误进行详细诊断，以便学生知道如何纠正错误。事实上，许多商业设施发现需要修改程序，因此重新编译的频率比 NASA 的风洞经验建议的要高。这些设施也发现需要缩减围绕大型机设施发展起来的批处理计算操作。

In a teaching environment the opposite was the case: the programs were bound to contain errors. Because the programs were short and typically did not handle large quantities of data, it mattered less that the compiled program execute quickly. Since many students would be submitting different programs, it did matter that the compiler work quickly, and that it produce, besides machine code, a detailed diagnosis of the errors it found, so that the student would know how to correct them. In fact, many commercial installations found a need to tinker with a program and thus recompile more frequently than NASA’s wind tunnel experience would suggest. These, too, found a need to telescope the operations of batch computing that had grown up around mainframe installations.

计算机的批处理方法在许多大学中仍然占据着中心地位，原因无他，那就是大学希望学生毕业后熟悉工业世界中的系统。几个大学部门开发了可以编译学生程序并立即指示机器执行的系统。这些“加载即用”编译器使学生能够获得更快的响应。84 ^（第 3 章中描述的密歇根大学 MAD 系统是该系统的早期版本。）

The batch method of computer use remained at center stage at many universities, if for no other reason than that universities wanted their students to become familiar with the kind of systems that they would find in the industrial world upon graduation. Several university departments developed systems that would compile a student’s program, and immediately direct the machine to execute it. These “load-and-go” compilers allowed the student to get a much quicker response.⁸⁴ (The University of Michigan MAD system, described in chapter 3, was an early version of this.)

将批处理计算应用于教学的最具创新性的工作是在加拿大滑铁卢大学完成的。滑铁卢大学的计算机科学系是最早成立的系之一（1962 年），不久之后，在 J. Wesley Graham 的领导下，该系的计算机教学本科生数量几乎超过了加拿大其他任何学校。1967 年，该系显然还拥有全加拿大最强大的计算机（System/360 Model 75）。到那一年，该系已经为另一台计算机开发了一个 FORTRAN 编译器。它被称为 WATFOR（Waterloo FORTRAN），基于威斯康星大学在 20 世纪 60 年代初开发的类似编译器：

The most innovative work in adapting batch computing to teaching was done at the University of Waterloo Canada. Waterloo’s computer science department was among the first to be founded (in 1962), and before long, under the leadership of J. Wesley Graham, it was teaching computing to more undergraduates than almost any other school in Canada. In 1967 it apparently also owned the most powerful computer in all of Canada (a System/360 Model 75). By that year the department already had developed a FORTRAN compiler for another computer. It was called WATFOR (Waterloo FORTRAN) and was based on a similar compiler developed at the University of Wisconsin in the early 1960s:

WATFOR 是由四名三年级数学系学生于 1965 年夏天编写的。它是一款快速的内核编译器，具有良好的错误诊断功能，这对学生调试程序以及加快执行速度特别有用^。85

WATFOR was written by four third-year math students in the summer of 1965. It was a fast in-core compiler with good error diagnostics, which proved especially useful to students for debugging their programs, as well as speeding up execution.⁸⁵

该编译器使大学计算机每小时可以运行 6,000 个作业，而且根据 Graham 的估计，运行一个学生程序的成本从 10 美元降低到了 10 美分。86 ^WATFOR在 360 Model 75 问世后被重写；它也进行了升级（并被赋予了古怪的名字 WATFIV）。Graham 的教科书《包含 WATFOR 和 WATFIV 的 Fortran IV》影响了一代计算机科学专业的学生，他们从中学习编程。87^该大学于 1972 年为 360/75 开发了类似的 COBOL 语言编译器 WATBOL；它还开发了 Waterloo SCRIPT，这是一个在独立文字处理器罕见的时代广泛使用的文本处理程序。滑铁卢大学将该软件分发给世界各地的学术计算中心，通过适度的服务费为大学带来了稳定的收入。

The compiler allowed the university computer to run 6,000 jobs an hour, and by Graham’s estimate reduced the cost of running a student’s program from ten dollars to ten cents.⁸⁶ WATFOR was rewritten for the 360 Model 75 when it arrived; it was also upgraded (and given the whimsical name WATFIV). Graham’s textbook Fortran IV with WATFOR and WATFIV influenced a generation of computer science students, who learned programming from it.⁸⁷ The university developed a similar compiler for the COBOL language, called WATBOL, for the 360/75 in 1972; it also developed Waterloo SCRIPT, a text-processing program widely used at a time when stand-alone word processors were rare. Waterloo distributed this software to academic computing centers worldwide, earning the university a steady stream of revenue from modest service fees.

达特茅斯学院的 BASIC

BASIC at Dartmouth

分时为大学教学提供了另一种途径。可以设置分时环境来处理来自许多用户的小程序和数据集，让每个用户都能快速诊断出他们可能犯的任何错误。仅仅拥有分时是不够的；还必须设计系统，以便用户可以轻松编写程序并快速、清晰地诊断任何错误。在数学系主任（后来成为学院院长）约翰·G·凯梅尼 (John G. Kemeny) 的领导下，达特茅斯学院于 1963 年开始建立这样的系统。凯梅尼曾使用穿孔卡设备为洛斯阿拉莫斯的核武器设计进行计算。那段经历使他相信“除了最初开发通用高速计算机之外，最重要的事件是人机交互的到来。” ⁸⁸他想要一个可以向达特茅斯所有学生（而不仅仅是那些学习计算机科学、工程或物理的学生）教授交互式计算的系统。他知道剑桥地区正在进行的工作，包括基于 IBM 的 CTSS 和在 DEC PDP-1 上运行的系统。89^麻省理工学院从这些不起眼的起点开始，开始更雄心勃勃的 MAC 项目，而 Kemeny 和 Thomas E. Kurtz（也是达特茅斯数学系的）决定围绕一种专门针对达特茅斯学生需求的编程语言构建一个适度的系统。他们称该语言为 BASIC。

Time-sharing offered another avenue for university instruction. A timesharing environment could be set up to handle small programs and data sets from many users, giving each a rapid diagnosis of any errors he or she might have made. It was not enough simply to have time-sharing; one also had to design the system so that users could write programs easily and receive a quick and intelligible diagnosis of any errors. Under the leadership of John G. Kemeny, chairman of the mathematics department (and later president of the college), Dartmouth began building such a system in 1963. Kemeny had done calculations using punched-card equipment for the design of atomic weapons at Los Alamos. That experience led him to believe that “next to the original development of general-purpose high-speed computers the most important event was the coming of man-machine interaction.”⁸⁸ He wanted a system that would teach interactive computing to all of Dartmouth’s students—not just those studying computer science or engineering or physics. He was aware of work being done in the Cambridge area, including the IBM-based CTSS and a system running on a DEC PDP-1.⁸⁹ Whereas MIT went on from these modest beginnings to the more ambitious Project MAC, Kemeny and Thomas E. Kurtz (also of the Dartmouth mathematics department) decided to build a modest system around a programming language especially tailored to the needs of Dartmouth students. They called that language BASIC.

贝尔实验室和 MAC 项目选择了通用电气计算机作为其雄心勃勃的分时系统，现在达特茅斯也选择了通用电气。达特茅斯使用通用电气 235 计算机连接到较小的通用电气数据网络计算机，后者收集和管理分散在校园各处的电传打字机终端的信号。达特茅斯的系统不是约翰·麦卡锡等支持者定义的通用分时系统。它专注于本科教育，而学校只有 25% 的学生主修科学或工程学。麻省理工学院希望其系统支持多种语言；在达特茅斯，学生只用 BASIC 编程^。90

Bell Laboratories and Project MAC had chosen General Electric computers for their ambitious time-sharing systems, and now Dartmouth, too, chose GE. Dartmouth used a General Electric 235 computer connected to a smaller GE Datanet computer, which collected and managed the signals from the Teletype terminals scattered across the campus. Dartmouth’s system was not a general-purpose time-sharing system, as defined by proponents like John McCarthy. It was tightly focused on undergraduate education, at a school where only 25 percent of students majored in science or engineering. MIT wanted a variety of languages available on its systems; at Dartmouth, the students programmed only in BASIC.⁹⁰

达特茅斯的经验对达特茅斯和通用电气来说都是成功的。但是达特茅斯的开放、免费访问模式（如大学图书馆）并没有盛行。通用电气提供了商业版本，其他几所大学也采用了该模式，在通用电气将其计算机业务出售给霍尼韦尔后，这些大学继续得到支持。大型计算机仍然很昂贵，对达特茅斯这样的小型私立大学有效的方法不一定适用于拥有大量多元化研究生和本科生的州立大学。大多数大学认为有必要根据学生连接到计算机系统的时间收取费用，并且其程序使用中央处理器的时间越长，费用就越高。实际支付这笔费用的人因大学而异，但通常涉及国家科学基金会或其他政府机构。91^在许多情况下，支付的钱很少。计算机制造商给大学提供硬件折扣；它还可能申请减税；大学想方设法将剩余的费用埋入其他联邦资助（有时是军事）的研究项目中。许多大学继续教授使用穿孔卡、Fortran、COBOL 和批处理的计算方法。

The Dartmouth experience was a success for both Dartmouth and GE. But Dartmouth’s model of open, free access, like the college library, did not prevail. General Electric offered a commercial version, and a few other universities adopted the model, and these continued to be supported after General Electric sold its computer business to Honey-well. Mainframe computers were still expensive, and what worked for a small, private college like Dartmouth did not necessarily work at a state university with a large, diverse graduate and undergraduate student body. Most universities felt a need to charge a fee based on the time a student was connected to the computer system, with higher fees charged for the time his or her program used the central processor. Who actually paid this fee varied from one university to the next, although often the National Science Foundation or some other government agency was involved.⁹¹ In many cases, little real money was paid. The computer manufacturer gave the university a discount on the hardware; it may also have claimed a tax deduction; the university found ways to bury the remaining charges into some other, federally funded (sometimes military) research project. Many universities continued to teach computing using punched cards, Fortran, COBOL, and batch processing.

尽管达特茅斯模式的影响力不大，但他们开发的编程语言 BASIC 却成为世界上使用最广泛的计算机编程语言之一，其影响力远远超出了它所针对的分时系统。我们看到了 IBM System/360 如何在保留其基本结构的同时不断发展数十年；BASIC 也在不断发展以服务于新市场，同时保持其易用性。它最终成为一种推动个人电脑进入主流的语言，与为个人电脑提供最佳 BASIC 的公司——微软公司一起。

Although the Dartmouth model had only a modest influence, the programming language they developed, BASIC, became one of the most widely used computer programming languages in the world with an influence that extended well beyond the time-sharing system for which it was written. We saw how the IBM System/360 was able to evolve for decades while retaining its essential structure; so too did BASIC evolve to serve new markets while preserving its ease of use. It eventually became a language that propelled the personal computer into the mainstream, along with the company that provided the best BASIC for personal computers, the Microsoft Corporation.

1971 年，数字设备公司迈出了 BASIC 演进的关键一步。针对刚刚发布的 PDP-11，DEC 开发了一个名为“资源共享分时” (RSTS-11) 的系统，允许在 PDP-11 上进行分时。最初，该系统用于最简单的 PDP-11 Model 20；后来的版本运行在更大的型号上。PDP-11/20 的功率仅为达特茅斯 GE 大型机的一小部分，其 56K 核心内存按照 1990 年代的标准被视为玩具。它没有硬件设施来保护内存位置免遭故意或意外更改——大多数人认为这是分时绝对必要的^。92

The crucial step in the evolution of BASIC was taken in 1971 at the Digital Equipment Corporation. For the just-announced PDP-11, DEC developed a system called “Resource Sharing Time Sharing” (RSTS-11) that allowed time-sharing on the PDP-11. Initially it was offered for the Model 20, the simplest PDP-11; later versions ran on bigger models. The PDP-11/20 had a fraction of the power of the GE mainframe at Dartmouth, and with its 56K core memory would be considered a toy by 1990s standards. It had no facilities in hardware for protecting memory locations from alteration, either deliberate or accidental—something most thought was absolutely necessary for time-sharing.⁹²

由 Mark Bramhall 领导的工程师团队完全用 BASIC 实现了 RSTS-11，但这是一个带有一些有趣扩展的 BASIC 版本。RSTS-11 需要进行系统调用来登录或注销等，这由名为“SYS”的命令实现。用户可以通过另一个名为 PEEK 的命令调用存储在内存特定位置的各个字节。具有特殊权限的 PDP-11 用户可以将字节直接 POKE 到内存中 - 与 PEEK 相反，尽管与 PEEK 不同，这是一个非常危险的命令，可能会破坏分时可行性。这些命令在达特茅斯使用的通用电气系统上都不可行，但对于 PDP-11 来说，它们可以工作，并且效果很好。

A team of engineers led by Mark Bramhall implemented RSTS-11 entirely in BASIC, but a version of BASIC with some interesting extensions. RSTS-11 needed to make system calls to log on or off and the like, which was implemented by a command called “SYS.” A user could recall the individual bytes stored at a specific location in memory by another command, called PEEK. A PDP-11 user with special privileges could POKE bytes directly into memory—the reverse of PEEK, although unlike PEEK, this was a very dangerous command that could destroy the viability of time-sharing. None of these commands would have been feasible on the General Electric system used at Dartmouth, but for the PDP-11 they worked, and worked well.

除了添加这些命令外，DEC 工程师还进一步修改了 BASIC，使其在不占用太多内存的情况下也能运行。这些实现严重损害了 Kemeny 和 Kurtz 关于语言应该是什么样子的一些原则，这让达特茅斯教授们很不高兴。93^但结果是，即使是简单的小型计算机安装也不再需要机器语言。DEC 的 BASIC 的这些功能组合——能够执行低级系统调用或字节传输，并且能够适应内存有限的机器——几年后被微软公司采用，用于其第一台个人电脑的 BASIC 版本。

Besides adding these commands, DEC engineers further modified BASIC so that it also could be implemented without taking up much memory. These implementations severely compromised some of Kemeny and Kurtz’s principles of what the language ought to look like, something that upset the Dartmouth professors.⁹³ But as a result, one no longer needed machine language for even a simple minicomputer installation. This combination of features of DEC’s BASIC—its ability to do low-level system calls or byte transfers, and its ability to fit on machines with limited memory—would be adopted by the Microsoft Corporation a few years later for its version of BASIC for the first personal computers.

基于 PDP-11/45 和惠普 HP-2000 等新型小型计算机的更先进型号的分时系统非常流行。这些系统为计算机科学系提供了大型机的替代方案。在 20 世纪 70 年代，随着这门学科的兴起，许多大学和学院发现他们可以以适中的成本围绕分时小型计算机建立一个不错的课程。对于初学者，有 BASIC；对于更高级的学生，有更高级的语言，比如 Pascal，当时它被认为是一种更好的教学语言。这种多样性弥补了这些学生没有体验使用 IBM System/370 的“大机器”及其庞大的 OS/MVS 操作系统进行数据处理世界的乐趣这一事实。在某些情况下，这些学生学到的东西远远超过使用大型机的学生所学到的东西——UNIX 操作系统，它是在 DEC 小型计算机上开发的。因此，这一代计算机造就了新一代的学生；学生们认为小型计算机的体积小、成本低、交互使用以及 UNIX 操作系统的强大功能是理所当然的。

Time-sharing systems based on more advanced models of the new minicomputers, like the PDP-11/45 and the Hewlett-Packard HP-2000, were very popular. These systems provided an alternative to the mainframe for computer science departments. Through the 1970s, as that discipline emerged, many universities and colleges found they could build a respectable curriculum around a time-shared minicomputer at a modest cost. For the beginning student there was BASIC; for those more advanced, there were more advanced languages, such as Pascal, then in favor as a better teaching language. This variety compensated for the fact that these students did not experience the flavor of the world of data processing using the “big iron” of the IBM System/370 and its giant OS/MVS operating system. In some cases, these students learned something far in advance of what students using mainframes learned—the UNIX operating system, which had been developed on DEC minicomputers. This generation of computers thus created a new generation of students; students who took for granted the small size, low cost, and interactive use of the minicomputer, and the power of the UNIX operating system.

到 20 世纪 70 年代中期，小型计算机已在多个市场中占据了强势地位，并已脱离了 OEM 市场嵌入式处理器的利基市场。大型机通过其通道在磁带驱动器和“磁盘场”之间来回传输大量数据的能力阻碍了小型计算机进入商业数据处理市场。但小型计算机比大型机更好地利用了集成电路、封装和处理器架构方面的进步。它的未来似乎确实一片光明。然而，接下来发生的事情并不是它的创造者想要的。小型计算机为来自完全不同来源的个人计算机铺平了道路，为它自己毁灭埋下了祸根。

By the mid-1970s, the minicomputer had established strong positions in several markets and had moved out of its niche as an embedded processor for the OEM market. What held it back from the business data-processing market was the mainframe’s ability to move enormous quantities of data through its channels, back and forth to rows of tape drives and “disk farms.” But the minicomputer took better advantage than mainframes of advances in integrated circuits, packaging, and processor architecture. Its future seemed bright indeed. What happened next, however, was not what its creators intended. The mini generated the seeds of its own destruction, by preparing the way for personal computers that came from an entirely different source.

个人电脑，1972-1977 年

The Personal Computer, 1972–1977

无论你是否做好准备，计算机都即将走进人们的生活。

这是个好消息，也许是自迷幻药问世以来最好的消息。

Ready or not, computers are coming to the people.

That’s good news, maybe the best since psychedelics.

这句话引出了《滚石》杂志¹五周年纪念刊（1972 年 12 月 7 日）的一篇文章。《太空战争：计算机迷们的狂热生活和象征性死亡》的作者是斯图尔特·布兰德，他是一位身材瘦高的加利福尼亚人，已经因出版《全球概览》而声名鹊起。即使对于一位公认的反主流文化英雄来说，布兰德的履历也是独一无二的。20 世纪 60 年代，在斯坦福大学，他曾参与国防部赞助的致幻药物实验。1968 年，他帮助道格·恩格尔巴特在旧金山举行的秋季联合计算机会议的一次现已成为传奇的会议上展示了他的交互式计算工作。2^布兰德对计算机以及将其用作交互式工具的新方法并不陌生。

Those words introduced a story in the fifth anniversary issue of Rolling Stone¹ (December 7, 1972). “Spacewar: Fanatic Life and Symbolic Death Among the Computer Bums” was written by Stewart Brand, a lanky Californian who had already made a name for himself as the publisher of the Whole Earth Catalog. Brand’s resume´ was unique, even for an acknowledged hero of the counterculture. At Stanford in the 1960s, he had participated in Defense Department–sponsored experiments with hallucinogenic drugs. In 1968 he had helped Doug Engelbart demonstrate his work on interactive computing at a now-legendary session of the Fall Joint Computer Conference in San Francisco.² Brand was no stranger to computers or to the novel ways one might employ them as interactive tools.

布兰德是对的。计算机确实进入了大众市场。计算机进入大众市场对社会的影响可能比精神类药物的传播更大。然而，个人计算机的出现方式却不像布兰德或几乎任何人想象的那样。个人计算机的发展轨迹很难用理性来解释。当试图描述从 1972 年到 1977 年的那几年时，人们会想起马克·吐温的话：“很少有事情会在正确的时间发生，其余的事情则根本不会发生。有责任心的历史学家会纠正这些缺陷。” ³本章将探讨计算机如何“进入大众”，不是像吐温笔下的历史学家所写的那样，而是像它真实发生的那样。

Brand was right. Computers did come to the people. The spread of computing to a mass market probably had a greater effect on society than the spread of mind-altering drugs. Personal computing, however, did not arrive in the way that Brand—or almost anyone else—thought it would. The development of personal computing followed a trajectory that is difficult to explain as rational. When trying to describe those years, from 1972 through 1977, one is reminded of Mark Twain’s words: “Very few things happen at the right time, and the rest do not happen at all. The conscientious historian will correct these defects.”³ This chapter will examine how computers came “to the people,” not as Twain’s historian would have written it, but as it really occurred.

激发布兰德灵感的，是他看到斯坦福人工智能实验室的人们玩电脑游戏《太空大战》。《太空大战》揭示了计算机与不折叠主轴或损坏打孔卡的环境相去甚远。他们使用的硬件不是“个人的”，但使用方式却是个人的：为了好玩，为了互动，而不关心处理器使用了多少次。两年后，当个人电脑突然进入市场时，人们想要的就是这个。

What triggered Brand’s insight was watching people at the Stanford Artificial Intelligence Laboratory playing a computer game, Spacewar. Spacewar revealed computing as far from the do-not-fold-spindle-or-mutilate punched-card environment as one could possibly find. The hardware they were using was not “personal,” but the way it was being used was personal: for fun, interactively, with no concern for how many ticks of the processor one was using. That was what people wanted when, two years later, personal computers burst into the market.

Spacewar 在 PDP-10 上运行。就硬件而言，PDP-10 与随后几十年的个人计算机没有任何共同之处。4^它很大——甚至 DEC 自己的文献也称它为大型机。5^它有 36 位字长。一套完整的系统成本约为 50 万美元，很容易就占据了一个房间。它使用分立晶体管和磁芯，而不是集成电路作为逻辑和内存。6^尽管如此，人们仍然可以将 PDP-10 视为个人计算机的祖先。它从一开始就被设计为支持交互式使用。虽然它的分时能力不如麻省理工学院的 MAC 项目那么雄心勃勃，但它运行良好。在所有早期的分时系统中，PDP-10 最能创造一种幻觉，即每个用户都得到了计算机的全部关注和资源。反过来，这种幻觉又创造了计算可能是什么的心理模型——这种心理模型后来在真正的个人计算机中得以实现^。7

Spacewar was running on a PDP-10. In terms of its hardware, a PDP-10 had nothing in common with the personal computers of the next decades.⁴ It was large—even DEC’s own literature called it a mainframe.⁵ It had a 36-bit word length. A full system cost around a half million dollars and easily took up a room of its own. It used discrete transistors and magnetic cores, not integrated circuits, for logic and memory. ⁶ Still, one can think of the PDP-10 as an ancestor of the personal computer. It was designed from the start to support interactive use. Although its time-sharing abilities were not as ambitious as those of MIT’s Project MAC, it worked well. Of all the early time-sharing systems, the PDP-10 best created an illusion that each user was being given the full attention and resources of the computer. That illusion, in turn, created a mental model of what computing could be—a mental model that would later be realized in genuine personal computers.⁷

第 5 章讨论了分时技术的早期发展，以及麻省理工学院 MAC 项目选择通用电气计算机。与此同时，麻省理工学院人工智能实验室获得了一台 DEC PDP-6（PDP-10 的前身）用于研究（图 7.1）。据传，麻省理工学院的学生，尤其是技术模型铁路俱乐部的成员，与 DEC 密切合作开发 PDP-6，特别是为其开发操作系统，这后来对 PDP-10 的系统软件产生了影响。8^为了讽刺附近 IBM 大型机上运行的兼容分时系统，学生们将他们的 PDP-6 系统称为 ITS（不兼容分时系统）。9 ^PDP -6 没有使其成为可行的分时系统所需的磁盘存储空间，因此只售出了大约 20 台。PDP-10 有一个随机存取磁盘系统，允许用户直接访问自己的个人文件。¹⁰与其他 DEC 计算机一样，PDP-10 也允许用户将个人文件和程序加载到廉价的 DECtape 卷轴上，而这种卷轴可以轻松放入公文包中。

Chapter 5 discussed the early development of time-sharing and the selection of a General Electric computer for Project MAC at MIT. While that was going on, the MIT Artificial Intelligence Laboratory obtained a DEC PDP-6, the PDP-10’s immediate predecessor, for its research (figure 7.1). According to the folklore, MIT students, especially members of the Tech Model Railroad Club, worked closely with DEC on the PDP-6, especially in developing an operating system for it, which would later have an influence on the PDP-10’s system software.⁸ As a pun on the Compatible Time Sharing System that was running on an IBM mainframe nearby, the students called their PDP-6 system ITS—Incompatible Time Sharing System.⁹ The PDP-6 did not have the disk storage necessary to make it a viable time-sharing system and only about twenty were sold. The PDP-10 did have a random-access disk system, which allowed its users direct access to their own personal files.¹⁰ Like other DEC computers, the PDP-10 also allowed users to load personal files and programs onto inexpensive reels of DECtape, which fitted easily into a briefcase.

PDP-10 就像是个人电脑的感觉来自于它的操作系统——尤其是它管理磁盘或磁带之间信息流的方式。在麻省理工学院的帮助下，DEC 从 1972 年开始提供了一个名为“TOPS-10”的系统。在 TOPS-10 手册的介绍中，作者指出：“我们的目标一直是，在一个正确配置的系统中，每个用户在需要使用机器时，都感觉自己拥有机器的一部分。” ¹¹用户可以轻松地从终端创建、修改、存储和调用数据块。系统用已经熟悉的术语“文件”来称呼这些块。文件由一到六个字符命名，后跟一个句点，然后是三个字符的扩展名（通常表示文件的类型，例如：xxxxxx.BAS 表示用 BASIC 编写的程序）。通过在终端上输入 DIR，用户可以获得磁盘上所有文件的目录。他们可以轻松地将文件内容发送到所需的输出设备，该设备通常由三个字母的代码组成，例如，LPT 表示行式打印机，TTY 表示电传打字机^。12

The feeling that a PDP-10 was one’s own personal computer came from its operating system—especially from the way it managed the flow of information to and from the disks or tapes. With MIT’s help, DEC supplied a system called “TOPS-10,” beginning in 1972. In the introduction to the TOPS-10 manual, the authors stated, “Our goal has always been that in a properly configured system, each user has the feeling that he owns his portion of the machine for the time he needs to use it.”¹¹ Users could easily create, modify, store, and recall blocks of data from a terminal. The system called these blocks by the already-familiar term, “files.” Files were named by one to six characters, followed by a period, then a three-character extension (which typically told what type of file it was, e.g.: xxxxxx.BAS for a program written in BASIC). By typing DIR at a terminal users could obtain a directory of all the files residing on a disk. They could easily send the contents of a file to a desired output device, which typically consisted of a three-letter code, for example, LPT for line printer, or TTY for Teletype.¹²

图 7.1

有史以来最具影响力的计算机之一 DEC PDP-6，其创造者站在 Mill 两侧，1964 年。左边是 C. Gordon Bell，穿着运动夹克。PDP-6 的销量不佳，但却是更成功的 PDP-10 和 DEC System-20 的原型。它对计算进程的影响与当时推出的更著名的 PDP-8 一样大。（来源：数字设备公司。）

Figure 7.1

One of the most influential computers of all time, the DEC PDP-6, flanked by its creators at the Mill, 1964. C. Gordon Bell is at the left, wearing the sports jacket. The PDP-6 did not sell well but was the prototype for the more successful PDP-10 and DEC System-20. It would have as much of an impact on the course of computing as the much more celebrated PDP-8, also introduced at that time. (Source: Digital Equipment Corporation.)

TOPS-10 的一小部分始终存在于核心内存中。其他程序存储在磁盘上，可根据需要调用。一个名为 PIP（外围设备交换程序）的程序允许用户以各种方式将文件移入和移出输入/输出设备。另一个程序 TECO（文本编辑器和校正器）允许用户从终端编辑和操作文本。DDT（动态调试工具）允许用户分析程序并纠正错误，而无需经历批处理中漫长的周转时间。

A small portion of TOPS-10 was always present in core memory. Other programs were stored on the disk and could be called up as necessary. One, called PIP (Peripheral Interchange Program), allowed users to move files in a variety of ways to and from input/output equipment. Another program, TECO (Text Editor and Corrector), allowed users to edit and manipulate text from a terminal. DDT (Dynamic Debugging Tool) allowed users to analyze programs and correct errors without going through the long turnaround times that plagued batch processing.

对于 PDP-10 用户来说，TOPS-10 是简洁和优雅的奇迹，让他们产生一种可以自己控制的幻觉。TOPS-10 就像一辆大众甲壳虫：基本、简单、易于理解和操作。13^使用PDP-10 不仅有趣，而且令人上瘾。布兰德看到人们在一台电脑上玩太空大战，或者说冒险游戏（也许是所有电脑游戏中最持久的）也是在这台电脑上编写的，这绝非偶然^。14

For PDP-10 users, TOPS-10 was a marvel of simplicity and elegance and gave them the illusion that they were in personal control. TOPS-10 was like a Volkswagen Beetle: basic, simple, and easy to understand and work with.¹³ Using a PDP-10 was not only fun but addictive. It was no accident that Brand saw people playing Spacewar on one, or that it was also the computer on which Adventure—perhaps the most long-lasting of all computer games—was written.¹⁴

在西海岸，出现了另一种具有类似功能的系统，即南加州科学数据系统 (SDS) 提供的 SDS-940。940 是传统计算机 SDS 930 的扩展，由伯克利的研究人员在国防部高级研究计划局的支持下进行了改进。940 比 PDP-10 更精致，性能也更好。不过，PDP-10 似乎更受青睐。在施乐帕洛阿尔托研究中心，这个传奇实验室将创造大量个人计算机，鼓励工作人员使用 SDS 机器，因为施乐刚刚购买了 SDS。但那里的研究人员拒绝了，而是构建了一台 PDP-10 的克隆版，他们称之为 MAXC（多路访问施乐计算机），这个名字是对 SDS 创始人马克斯·帕列夫斯基的双关语。¹⁵（帕列夫斯基将 SDS 出售给施乐公司后变得非常富有，之后涉足好莱坞电影、政治和文化领域，并加入了《滚石》杂志的董事会。帕列夫斯基还利用这笔钱成为了一名风险投资家，帮助资助了英特尔等公司。）¹⁶

On the West Coast another system appeared with similar capabilities, the SDS-940, offered by Scientific Data Systems (SDS) of southern California. The 940 was an extension of a conventional computer, the SDS 930, modified by researchers at Berkeley with support from the Defense Department’s Advanced Research Projects Agency. The 940 was more polished than the PDP-10, and it performed well. Still, the PDP-10 seemed to be preferred. At the Xerox Palo Alto Research Center, the legendary lab where so much of personal computing would be created, the staff was encouraged to use SDS machines, since Xerox had just purchased SDS. But the researchers there resisted and instead built a clone of a PDP-10, which they called MAXC—Multiple Access Xerox Computer—the name a pun on Max Palevsky, the founder of SDS.¹⁵ (Palevsky, after becoming very wealthy from the sale of SDS to Xerox, dabbled in Hollywood movies, politics, and culture—and joined the board of Rolling Stone. Palevsky also became a venture capitalist with that money, helping to fund Intel, among other companies.)¹⁶

有一段时间，当华尔街迷恋与计算机相关的任何事物时，很容易筹集资金购买或租赁 PDP-10 或 SDS-940，然后将计算机时间出售给工程公司或其他客户。这些公司中的大多数都资金不足，不了解他们所销售产品的复杂性。与电力公用事业行业的同行一样，他们必须拥有足够的容量来处理峰值负载，以免让客户失望。但这意味着在非高峰时段，他们会浪费未使用的昂贵计算设备。管理业务周期所需的资本要求与电力业务一样大，电力业务经历了数十年的混乱和动荡才稳定下来。只有少数公司幸存下来，^{17 家}，甚至更少，如加利福尼亚州库比蒂诺的 Tymshare 表现良好（尽管它在 20 世纪 80 年代初被卖给了麦克唐纳-道格拉斯）。¹⁸在这些公司中，值得一提的是 Computer Center Corporation（简称 C-Cubed），该公司于 1968 年在西雅图地区安装了第一批 PDP-10。在创业初期，该公司为当地青少年比尔·盖茨提供了使用计算机的自由时间，以换取他帮助查找和清除系统中的错误。C-Cubed 于 1970 年倒闭，让盖茨尝到了交互式计算的潜力^。19

For a while, when Wall Street was enamored of anything connected with computers, it was easy to raise money to buy or lease a PDP-10 or SDS-940, and then sell computer time to engineering companies or other customers. Most of these firms were undercapitalized and did not understand the complexities of what they were selling. Like their counterparts in the electric utility industry, they had to have enough capacity to handle peak loads, in order not to discourage customers. But that meant that during off-peak times they would be wasting unused and expensive computing equipment. The capital requirements necessary to manage the cycles of the business were as large as they were in the electric power business, which had gone through decades of chaos and turmoil before settling down. Only a few survived,¹⁷ and even fewer, like Tymshare of Cupertino, California, did well (although it was sold to McDonnell-Douglas in the early 1980s).¹⁸ Among those many companies, one is worth mentioning, Computer Center Corporation, or C-Cubed, which installed one of the first PDP-10s in the Seattle area in 1968. While it was getting started, it offered a local teenager, Bill Gates, free time on the computer in exchange for helping find and rid the system of bugs. C-Cubed folded in 1970, having given Gates a taste of the potential of interactive computing.¹⁹

许多能够使用这些系统的人看到了计算的未来。但分时公司的财务困境也表明，要广泛普及个人互动使用将非常困难。曾有人尝试免费或低价向公众提供终端——最著名的是旧金山湾区的 Resource One 项目（部分资金来自《全球概览》）。但它也没有持续下去^。20

Many of those who had access to these systems saw the future of computing. But the financial troubles of time-sharing companies also showed that it would be difficult to make personal, interactive use widely available. There were attempts to make terminals accessible to the public for free or at low cost—the most famous being the Resource One project in the San Francisco Bay area (partially funded by the Whole Earth Catalog). But it did not last, either.²⁰

计算器和企业个人电脑项目

Calculators and Corporate Personal Computer Projects

经济因素阻碍了计算从上而下（从大型主机到分时终端）向公众的普及。但在这些尝试进行的同时，底层技术也在迅速发展。个人计算能否自下而上（从半导体电子技术的进步）实现？

Economics prevented the spread of computing to the public from the top down—from large mainframes through time-shared terminals. But while those attempts were underway, the underlying technology was advancing rapidly. Could personal computing arrive from the bottom up—from advances in semiconductor electronics?

许多工程师认为个人电脑的思维模型无关紧要。他们认为，个人电脑不是由任何人发明的，它只是半导体技术的进步而产生的。参与早期惠普计算器开发的工程师查克·豪斯 (Chuck House) 表示：“可以说我们基本上什么都没发明；我们只是采纳了所有现有的创意，并想出了如何以经济高效的方式实现它们。”戈登·贝尔 (Gordon Bell) 表示：“半导体密度确实是驱动力，随着密度水平的提高，不同的机器会随着时间的推移而出现。” ²¹对他们来说，发明就像是成熟后掉落到地上的水果，而发明者所做的只是捡起它。如果这是真的，那么人们会发现，随着半导体技术的进步，越来越多的机器能够提供个性化、互动式的使用。而这些机器将来自拥有工程和制造资源的知名公司，这些公司可以将这些进步转化为产品。

Many engineers believe that a mental model of the personal computer was irrelevant. They believe that no one invented the personal computer, it simply flowed from advances in semiconductors. Chuck House, an engineer involved with the early Hewlett-Packard calculators, said, “One could uncharitably say that we invented essentially nothing; we simply took all the ideas that were out there and figured out how to implement them cost-effectively.” Gordon Bell stated, “The semiconductor density has really been the driving force, and as you reach different density levels, different machines pop out of that in time.”²¹ To them, inventions are like a piece of fruit that falls to the ground when it is ripe, and the inventor is given credit for doing little more than picking it up. If that were true, one would find a steady progression of machines offering personal, interactive use, as advances in semiconductors made them viable. And these would have come from established firms who had the engineering and manufacturing resources to translate those advances into products.

利用半导体技术进步的产品确实出现在市场上。值得研究的是，它们是否证实或反驳了自下而上的个人电脑发明解释。

Products that took advantage of advances in semiconductors did appear on the market. It is worth looking at them to see whether they validate or refute the bottom-up explanation of the PC’s invention.

当然，第一台电子计算机的操作方式与个人计算机一样。一旦某人被授予使用机器的权限（在排队等待之后），他或她就可以使用整台计算机，无论出于何种目的。这让位于更受限制的访问，但麻省理工学院和林肯实验室的那些以这种方式使用 Whirlwind、TX-0 和 TX-2 的人永远不会忘记它的优势。1962 年，他们中的一些人开发了一台名为 LINC 的计算机，它由数字设备公司的逻辑模块组成，旨在供研究人员用作个人工具。由 NIH 资助的一项演示项目为生物医学研究人员提供了 16 台 LINC。DEC 生产了商业版本，到 20 世纪 60 年代末，大约有 1,200 台被用作个人计算机。LINC 的一个主要特点是其紧凑的磁带驱动器和磁带，人们可以轻松随身携带：这是 DECtape 的前身。磁带上数据的获取便利性与大型机中磁带的笨拙访问截然不同，这种便利性在个人电脑引入软盘系统后得以重现。22 ^DEC还推出了一款结合了 LINC 和 PDP-8 的计算机，售价 43,000 美元。尽管 DECtape 很快就出现在几乎所有 DEC 产品上，但 LINC 并没有取得与 PDP-8 和 PDP-11 系列小型计算机相同的商业成功^。23

The first electronic computers were of course operated as if they were personal computers. Once a person was granted access to a machine (after literally waiting in a queue), he or she had the whole computer to use, for whatever purpose. That gave way to more restricted access, but those at MIT and Lincoln Labs who used the Whirlwind, TX-0, and TX-2 that way never forgot its advantages. In 1962 some of them developed a computer called the LINC, made of Digital Equipment Corporation logic modules and intended for use by a researcher as a personal tool. A demonstration project, funded by the NIH, made sixteen LINCs available to biomedical researchers. DEC produced commercial versions, and by the late 1960s, about 1,200 were in use as personal computers. A key feature of the LINC was its compact tape drive and tapes that one could easily carry around: the forerunner of DECtape. The ease of getting at data on the tape was radically different from the clumsy access of tape in mainframes, and this ease would be repeated with the introduction of floppy-disk systems on personal computers.²² DEC also marketed a computer that was a combination of a LINC and a PDP-8, for $43,000. Although DECtape soon was offered on nearly all DEC’s products, the LINC did not achieve the same kind of commercial success as the PDP-8 and PDP-11 lines of minicomputers.²³

芯片密度的提高首先对个人设备中的计算器产生了影响。24^几十年来，能够执行四则运算加平方根功能的机器一直市场很小。20 世纪 50 和 60 年代，计算器行业由美国的 Friden 和 Marchant 以及欧洲的 Odhner 等公司主导。它们的产品复杂、笨重且价格昂贵。25 ¹⁹⁶⁴年，曾在哈佛大学与霍华德·艾肯共事的中国移民王安创办了王安实验室，推出了一款电子计算器。王安 LOCI 功能比最好的机械机器更多，但成本更低。它的继任者王安 300 甚至更易于使用且价格更便宜，部分原因是王安故意将 300 的价格定得低于 Friden 和其他公司的机械计算器。²⁶（只有一两家机械计算器公司在向电子计算器的过渡中幸存下来。）几年后，以示波器和电子测试设备闻名的惠普公司推出了 HP-9100A，这是一款售价不到 5,000 美元的计算器。意大利公司 Olivetti 推出了 Programma 101，这是一款售价 3,500 美元的计算器，主要用于会计和统计工作。除了直接计算，这些机器还可以执行记录在磁卡上的一小段步骤。27与^LINC一样，这些计算器使用分立电路。为了显示数字，王氏使用了“Nixie”管，这是 Burroughs 于 1957 年发明的一种巧妙的管子。惠普使用小型阴极射线管，这对于一家制造示波器的公司来说是意料之中的事。

Advances in chip density first made an impact on personal devices in calculators.²⁴ For decades there had been a small market for machines that could perform the four functions of arithmetic, plus square root. In the 1950s and 1960s the calculator industry was dominated by firms such as Friden and Marchant in the United States, and Odhner in Europe. Their products were complex, heavy, and expensive.²⁵ In 1964 Wang Laboratories, a company founded by An Wang, a Chinese immigrant who had worked with Howard Aiken at Harvard, came out with an electronic calculator. The Wang LOCI offered more functions, at a lower cost, than the best mechanical machines. Its successor, the Wang 300, was even easier to use and cheaper, partly because Wang deliberately set the price of the 300 to undercut the competitive mechanical calculators from Friden and others.²⁶ (Only one or two of the mechanical calculator firms survived the transition to electronics.) A few years later Hewlett-Packard, known for its oscilloscopes and electronic test equipment, came out with the HP-9100A, a calculator selling for just under $5,000. And the Italian firm Olivetti came out with the Programma 101, a $3,500 calculator intended primarily for accounting and statistical work. Besides direct calculation, these machines could also execute a short sequence of steps recorded on magnetic cards.²⁷ Like the LINC, these calculators used discrete circuits. To display digits, the Wang used “Nixie” tubes, an ingenious tube invented by Burroughs in 1957. HP used a small cathode-ray tube, as might be expected from a company that made oscilloscopes.

1970 年，第一款使用集成电路的计算器问世，价格便宜且体积小得多。28^它们大小与一本平装书相当，价格低至 400 美元。许多富有的消费者立即购买了它们，但直到 Bowmar 在 1971 年圣诞节期间以低于 250 美元的价格宣传 Bowmar Brain 后，计算器才进入公众视野。29^价格暴跌：1972 年低于 150 美元；1973 年低于 100 美元；1976 年低于 50 美元；最后它们变得足够便宜，可以作为促销小玩意儿免费赠送。30^与此同时，惠普在 1972 年初以 HP-35 震惊了市场，这是一款售价 400 美元的袖珍计算器，可以执行工程师和科学家所需的所有对数和三角函数。几年之内，计算尺就和机械计算器一起出现在博物馆的架子上^。31

By 1970 the first of a line of dramatically cheaper and smaller calculators appeared that used integrated circuits.²⁸ They were about the size of a paperback book and cost as little as $400. A number of wealthy consumers bought them immediately, but it wasn’t until Bowmar advertised a Bowmar Brain for less than $250 for the 1971 Christmas season that the calculator burst into public consciousness.²⁹ Prices plummeted: under $150 in 1972; under $100 by 1973; under $50 by 1976; finally they became cheap enough to be given away as promotional trinkets.³⁰ Meanwhile Hewlett-Packard stunned the market in early 1972 with the HP-35, a $400 pocket calculator that performed all the logarithmic and trigonometric functions required by engineers and scientists. Within a few years the slide rule joined the mechanical calculator on the shelves of museums.³¹

就像加工食品的成本主要在于包装和营销一样，计算器也是如此：技术不再决定商业成功。两家拥有消费者营销技能的日本公司卡西欧和夏普很快就占据了主导地位。在价值 50 万美元的 ENIAC 完成 30 年后，数字设备成为了一次性商品。先驱计算器公司要么停止生产计算器（王安电子也是如此），要么破产（Bowmar 也是如此）。惠普通过专注于更先进、更昂贵的型号而幸存下来；德州仪器则通过削减成本而幸存下来。

Like processed foods, whose cost is mostly in the packaging and marketing, so with calculators: technology no longer determined commercial success. Two Japanese firms with consumer marketing skills, Casio and Sharp, soon dominated. Thirty years after the completion of the half-million dollar ENIAC, digital devices became throw-away commodities. The pioneering calculator companies either stopped making calculators, as did Wang, or went bankrupt, as did Bowmar. Hewlett-Packard survived by concentrating on more advanced and expensive models; Texas Instruments survived by cutting costs.

商品价格让人很容易忘记这些计算器是工程学上的巧妙杰作。有些计算器可以在内存中存储一系列按键，从而执行简短的程序。第一台可编程袖珍计算器是惠普的 HP-65，于 1974 年初推出，售价 795 美元（图 7.2）。德州仪器和其他公司很快也效仿了这种计算器。尽管它们功能强大，但行业媒体仍不愿称它们为计算机，即使惠普将 HP-65 作为“个人计算机”推出（这可能是该术语首次在印刷品中使用）。32^它们有限的编程能力被其内置的对数和三角函数计算能力以及精确到十进制数的浮点运算能力所抵消。很少有大型计算机可以在没有定制软件的情况下做到这一点。

The commodity prices make it easy to forget that these calculators were ingenious pieces of engineering. Some of them could store sequences of keystrokes in their memory and thus execute short programs. The first of the programmable pocket calculators was Hewlett-Packard’s HP-65, introduced in early 1974 for $795 (figure 7.2). Texas Instruments and others soon followed. As powerful as they were, the trade press was hesitant to call them computers, even if Hewlett-Packard introduced the HP-65 as a “personal computer” (possibly the first use of that term in print).³² Their limited programming was offset by their built-in ability to compute logarithms and trigonometric functions, and to use floating-point arithmetic to ten decimal digits of precision. Few mainframes could do that without custom-written software.

图 7.2

HP-65。（来源：史密森学会。）

Figure 7.2

HP-65. (Source: Smithsonian Institution.)

袖珍可编程计算器的推出对计算技术的发展方向产生了几个深远的影响。首先，计算器就像 20 世纪 60 年代的民兵计划和阿波罗计划一样，创造了一个市场，供应商可以依靠长期生产，从而获得规模经济和低价。随着芯片密度（从而是能力）的提高，芯片制造商面临着亨利·福特在生产 T 型车时遇到的同样问题：只有长期生产同一种产品才能降低价格，但市场并非一成不变。集成电路尤其如此，随着集成度的提高，集成电路的功能本质上变得越来越专业化。（唯一的例外是内存芯片，这也是英特尔成立时专注于内存的原因之一。）计算器为逻辑芯片提供了第一个消费市场，使公司能够分摊设计复杂集成电路的高成本。1971 年至 1976 年间计算器价格的大幅下降表明了这种力量有多么强大^。33

The introduction of pocket programmable calculators had several profound effects on the direction of computing technology. The first was that the calculator, like the Minuteman and Apollo programs of the 1960s, created a market where suppliers could count on a long production run, and thereby gain economies of scale and a low price. As chip density, and therefore capability, increased, chip manufacturers faced the same problem that Henry Ford had faced with his Model T: only long production runs of the same product led to low prices, but markets did not stay static. That was especially true of integrated circuits, which by nature became ever more specialized in their function as the levels of integration increased. (The only exception was in memory chips, which is one reason why Intel was founded to focus on memories.) The calculator offered the first consumer market for logic chips that allowed companies to amortize the high costs of designing complex integrated circuits. The dramatic drop in prices of calculators between 1971 and 1976 showed just how potent this force was.³³

第二个影响同样重要。袖珍计算器，尤其是可编程的计算器，释放了众多个人的创造力和能量。这种力量已经在麻省理工学院和斯坦福大学创造了黑客文化（至少有一位麻省理工学院的教授对此感到不安）。34^他们的故事是硬件和软件设计枯燥的技术叙述中比较丰富多彩的故事之一。他们和他们的成就，经过适当的修饰，成为大众媒体的热门话题。当然，他们奇怪的个人习惯可以成为一个好故事，但这些是真的吗？开发系统软件是一项艰苦的工作，一个在正常工作时间工作、晚上还要回家照顾家人的领薪员工不太可能做好。分时系统将所有用户从提交一副牌和等待打印输出的繁琐中解放出来，但它迫使一些用户在深夜工作，而此时分时系统负载较轻，因此响应速度更快。

The second effect was just as important. Pocket calculators, especially those that were programmable, unleashed the force of personal creativity and energy of masses of individuals. This force had already created the hacker culture at MIT and Stanford (observed with trepidation by at least one MIT professor).³⁴ Their story is one of the more colorful among the dry technical narratives of hardware and software design. They and their accomplishments, suitably embellished, have become favorite topics of the popular press. Of course their strange personal habits made a good story, but were they true? Developing system software was hard work, not likely to be done well by a salaried employee, working normal hours and with a family to go home to in the evening. Time-sharing freed all users from the tyranny of submitting decks of cards and waiting for a printout, but it forced some users to work late at night, when the time-shared systems were lightly loaded and thus more responsive.

声称黑客创造了现代交互式计算大约只对了一半。从数量上看，可能从来没有超过几百人有幸被允许“黑客”攻击 PDP-10 之类的计算机（即不做雇主指定的编程工作）。到 1975 年，使用的 HP-65 可编程计算器超过 25,000 台，每台都归个人所有，他们可以用它做任何想做的事情。35^这些人是谁？HP-65 的用户并不“奇怪”。几乎所有人都是成年职业男性，包括土木工程师和电气工程师、律师、金融人士、飞行员等等。只有少数是学生（或教授），因为 HP-65 的价格为 795 美元。大多数人购买 HP-65 是因为他们在工作中有实际的计算需求。但这是一台个人机器——晚上可以带回家。这些用户（可能占拥有计算机的人数的 5% 或 10%）并不符合人们普遍认为的“衣衫褴褛、脸色苍白、头发蓬乱”的黑客形象。36^但他们对于编程的热情让他们成为了技术模型铁路俱乐部学生的智力表亲。随着计算器价格的下降，他们的数量还在不断增加，这首次表明个人计算确实是一种大众现象。

The assertion that hackers created modern interactive computing is about half-right. In sheer numbers there may never have been more than a few hundred people fortunate enough to be allowed to “hack” (that is, not do a programming job specified by one’s employer) on a computer like the PDP-10. By 1975, there were over 25,000 HP-65 programmable calculators in use, each one owned by an individual who could do whatever he or she wished to with it.³⁵ Who were these people? HP-65 users were not “strange”. Nearly all were adult professional men, including civil and electrical engineers, lawyers, financial people, pilots, and so on. Only a few were students (or professors), because an HP-65 cost $795. Most purchased the HP-65 because they had a practical need for calculation in their jobs. But this was a personal machine—one could take it home at night. These users—perhaps 5 or 10 percent of those who owned machines—did not fit the popular notion of hackers as kids with “[t]heir rumpled clothes, their unwashed and unshaven faces, and their uncombed hair.”³⁶ But their passion for programming made them the intellectual cousins of the students in the Tech Model Railroad Club. And their numbers—only to increase as the prices of calculators dropped—were the first indication that personal computing was truly a mass phenomenon.

惠普和德州仪器对这些事件毫无准备。他们把机器当成商品来销售；他们负担不起一支销售队伍，无法指导客户完成复杂的学习过程，从而最大限度地利用机器。这正是 IBM 销售人员的本职工作——但他们销售的大型机价值数百万美元。计算器的设计足够简单易用，因此至少对于基本任务来说，学习过程是不必要的。出乎意料的是，这些客户想要做的事情更多。供应商几乎帮不上什么忙，他们只能求助于彼此。用户组、俱乐部、通讯和出版物激增。

Hewlett-Packard and Texas Instruments were unprepared for these events. They sold the machines as commodities; they could ill-afford a sales force that could walk a customer through the complex learning process needed to get the most out of one. That was what IBM salesmen were known for—but they sold multimillion dollar mainframes. Calculators were designed to be easy enough to use to make that unnecessary, at least for basic tasks. What was unexpected was how much more some of those customers wanted to do. Finding little help from the supplier, they turned to one another. Users groups, clubs, newsletters, and publications proliferated.

这种支持性基础设施对于个人计算的成功至关重要；在接下来的十年里，它成为了一个独立的行业。许多个人计算机的历史都强调了这一点；他们经常引用 20 世纪 70 年代中期在斯坦福校园附近举行的家酿计算机俱乐部的作用，认为它尤其重要。37计算器用户组也很重要，尽管原因不同。随着^Altair等原始的第一台个人计算机让位于更完整的系统，许多计算器拥有者也购买了其中一台。在俱乐部的通讯中，不断讨论每台计算机的优缺点——一台机器能够轻松评估复杂的数学表达式，另一台机器更原始，但可能能够做所有这些事情甚至更多。38^家酿计算机俱乐部没有典型的成员，尽管计算器的拥有者往往是专业人士，他们的工作需要在白天进行计算，晚上会考虑其他用途。他们中的许多人对 PC 着迷；同时，他们对计算机采取了一种“给我看”的态度。你能依靠一个吗？你能用它设计雷达天线吗？它能处理中等规模的邮件列表吗？个人电脑是一台严肃的机器吗？起初的答案是“还不行”，但在这个社区的强烈推动下，平衡逐渐发生了变化。像家酿计算机俱乐部这样的团体强调个人电脑中的“个人”一词；计算器用户强调计算机这个词。

This supporting infrastructure was critical to the success of personal computing; in the following decade it would become an industry all its own. Many histories of the personal computer emphasize this point; they often cite the role of the Homebrew Computer Club, which met near the Stanford campus in the mid-1970s, as especially important.³⁷ The calculator users groups were also important, though for different reasons. As the primitive first personal computers like the Altair gave way to more complete systems, a number of calculator owners purchased one of them as well. In the club newsletters there were continuous discussions of the advantages and drawbacks of each—the one machine having the ability to evaluate complex mathematical expressions with ease, the other more primitive but potentially capable of doing all that and more.³⁸ There was no such thing as a typical member of the Homebrew Computer Club, although calculator owners tended to be professionals whose jobs required calculation during the day, and who thought of other uses at night. Many of them were bitten by the PC bug; at the same time they took a show-me attitude toward the computer. Could you rely on one? Could you use one to design a radar antenna? Could it handle a medium-sized mailing list? Was the personal computer a serious machine? At first the answers were, “not yet,” but gradually, with some firm prodding by this community, the balance shifted. Groups like the Homebrew Computer Club emphasized the “personal” in personal computer; calculator users emphasized the word computer.

自从分时技术和小型计算机揭示了大型计算机的替代方案以来，一直有先知和布道者对打孔卡和计算机房的世界大加批判，承诺创造一个真正交互式工具的数字天堂。其中最著名的是泰德·尼尔森，他自费出版的书《计算机自由》宣称（封面上还有一只举起的拳头）：“你现在就可以而且必须了解计算机。” ³⁹到 1974 年，这些梦想已经足够成为现实，以至于必须面对真正的“梦想机器”（尼尔森书的另一个标题）的具体能力和局限性。包括尼尔森在内的一些梦想家无法实现这一转变。他们拒绝了袖珍计算器。他们认为它太小、太便宜、不能做图形、不是“冯·诺依曼机器”等等。40^对他们来说，梦想机器更好，即使（或因为）它尚未建成。⁴¹到 1985 年，普通人的办公室和家中将有数百万台 IBM 个人电脑及其复制品。这些电脑将使用为其他目的开发的处理器，并几乎是偶然地适应个人电脑。但它们是真实的，是许多使用它们的人的灵感和创造力的源泉，同时也是那些知道它们可以有多好的人的挫败感之源。

Ever since time-sharing and minicomputers revealed an alternative to mainframe computing, there have been prophets and evangelists who raged against the world of punched cards and computer rooms, promising a digital paradise of truly interactive tools. The most famous was Ted Nelson, whose self-published book Computer Lib proclaimed (with a raised fist on the cover): “You can and must understand computers now.”³⁹ By 1974 enough of these dreams had become real that the specific abilities—and limits—of actual “dream machines” (the alternate title to Nelson’s book) had to be faced. Some of the dreamers, including Nelson, were unable to make the transition. They dismissed the pocket calculator. They thought it was puny, too cheap, couldn’t do graphics, wasn’t a “von Neumann machine,” and so on.⁴⁰ For them, the dream machine was better, even if (or because) it was unbuilt.⁴¹ By 1985 there would be millions of IBM Personal Computers and their copies in the offices and homes of ordinary people. These computers would use a processor that was developed for other purposes, and adapted for the personal computer almost by accident. But they would be real and a constant source of inspiration and creativity to many who used them, as well as an equal source of frustration for those who knew how much better they could be.

微处理器

The Microprocessor

计算器展示了集成电路的功能，但并未直接为个人交互式计算开辟道路。计算器中使用的芯片过于专业化，无法用于数值计算，因此无法成为通用计算机的基础。它们的架构是临时的，并受到各制造商的严密保护。所需要的是一组集成电路（甚至单个集成电路），其中包含通用存储程序计算机的基本架构。42^这种称为“微处理器”的芯片确实出现了。

Calculators showed what integrated circuits could do, but they did not open up a direct avenue to personal interactive computing. The chips used in them were too specialized for numerical calculation to form a basis for a general-purpose computer. Their architecture was ad-hoc and closely guarded by each manufacturer. What was needed was a set of integrated circuits—or even a single integrated circuit—that incorporated the basic architecture of a general-purpose, stored-program computer.⁴² Such a chip, called a “microprocessor,” did appear.

1964 年，当时在仙童公司工作、不久后成为英特尔联合创始人的戈登·摩尔 (Gordon Moore) 指出，自 1958 年发明集成电路以来，一块集成电路上可放置的电路数量每年都在翻一番。43^只需在一张半对数方格纸上绘制这个速率，“摩尔定律”便可预测，到 20 世纪 70 年代中期，人们可以买到一个芯片，其逻辑电路数量相当于 20 世纪 50 年代大型机中使用的逻辑电路。（回想一下，UNIVAC I 有大约 3,000 个电子管，与下面讨论的第一个微处理器中的有源元件数量大致相同。）到 20 世纪 60 年代末，晶体管-晶体管逻辑 (TTL) 已经成熟，但一种名为金属氧化物半导体 (MOS) 的新型半导体的出现，使人们可以在芯片上放置更多的逻辑元件。⁴⁴英特尔使用 MOS 生产其开创性的 1103 内存芯片，它是袖珍计算器成功的关键。MOS 允许的芯片密度使英特尔、德州仪器和其他半导体公司的工程师开始关注芯片计算机的概念。这并不意味着这种设备被认为是有用的。如果人们普遍知道芯片上可以放置足够的晶体管来制造计算机，那么人们也普遍认为这种芯片的市场如此之低，以至于其销售额永远无法收回所需的巨额开发成本^。45

In 1964 Gordon Moore, then of Fairchild and soon a cofounder of Intel, noted that from the time of the invention of integrated circuits in 1958, the number of circuits that one could place on a single integrated circuit was doubling every year.⁴³ By simply plotting this rate on a piece of semi-log graph paper, “Moore’s Law” predicted that by the mid 1970s one could buy a chip containing logic circuits equivalent to those used in a 1950s-era mainframe. (Recall that the UNIVAC I had about 3,000 tubes, about the same number of active elements contained in the first microprocessor discussed below.) By the late 1960s transistor-transistor logic (TTL) was well established, but a new type of semiconductor called metal-oxide semiconductor (MOS), emerged as a way to place even more logic elements on a chip.⁴⁴ MOS was used by Intel to produce its pioneering 1103 memory chip, and it was a key to the success of pocket calculators. The chip density permitted by MOS brought the concept of a computer-on-a-chip into focus among engineers at Intel, Texas Instruments, and other semiconductor firms. That did not mean that such a device was perceived as useful. If it was generally known that enough transistors could be placed on a chip to make a computer, it was also generally believed that the market for such a chip was so low that its sales would never recoup the large development costs required.⁴⁵

到 1971 年，这个想法在硅片上得以实现。几位工程师都为这项发明做出了贡献。英特尔工程师 Ted Hoff 提出了最初的概念，英特尔的 Federico Faggin 也因在硅片上实现这一概念而受到赞扬，德州仪器的 Gary Boone 也在那段时间设计了类似的电路。1990 年，在微处理器成为家喻户晓的商品多年后，经过多年的诉讼，来自加利福尼亚州拉帕尔马的独立发明家 Gil Hyatt 获得了专利。在法庭之外，他几乎没有支持者，最近的法庭裁决可能完全否定了他的主张^。46

By 1971 the idea was realized in silicon. Several engineers deserve credit for the invention. Ted Hoff, an engineer at Intel, was responsible for the initial concept, Federico Faggin of Intel deserves credit for its realization in silicon, and Gary Boone of Texas Instruments designed similar circuits around that time. In 1990, years after the microprocessor became a household commodity and after years of litigation, Gil Hyatt, an independent inventor from La Palma, California, received a patent on it. Outside the courts he has few supporters, and recent court rulings may have invalidated his claim entirely.⁴⁶

英特尔发明微处理器的故事已经被讲过很多次了。47^从本质上讲，这是一个以前就遇到过的故事：英特尔被要求为客户设计一个专用系统。它发现，通过设计一台通用计算机并使用软件来根据客户的需求进行定制，该产品将拥有更大的市场。

The story of the microprocessor’s invention at Intel has been told many times.⁴⁷ In essence, it is a story encountered before: Intel was asked to design a special-purpose system for a customer. It found that by designing a general-purpose computer and using software to tailor it to the customer’s needs, the product would have a larger market.

Intel 的这款电路的客户是 Busicom，这是一家日本公司，是手持计算器的顶级销售商。Busicom 试图生产一系列具有不同功能的产品，每种产品都针对不同的细分市场。它设想了一套定制设计的芯片，其中融入了高级数学函数的逻辑。Intel 管理层指派 1968 年加入公司的 Marcian E. (“Ted”) Hoff（Intel 的第十二名员工）与 Busicom 合作。

Intel’s customer for this circuit was Busicom, a Japanese company that was a top seller of hand-held calculators. Busicom sought to produce a line of products with different capabilities, each aimed at a different market segment. It envisioned a set of custom-designed chips that incorporated the logic for the advanced mathematical functions. Intel’s management assigned Marcian E. (“Ted”) Hoff, who had joined the company in 1968 (Intel’s twelfth employee), to work with Busicom.

英特尔的重点一直是半导体存储芯片。它一直回避 Busicom 建议的逻辑芯片，因为它认为这些芯片的市场有限。霍夫的洞察力在于认识到，通过设计更少的逻辑芯片和更通用的功能，人们可以完美地满足 Busicom 的需求。霍夫的灵感来自 PDP-8，它只有一组非常小的指令，但成千上万的用户已经对其进行了编程，使其可以执行各种操作。他还回忆起使用 IBM 1620，这是一台小型科学计算机，指令集极其有限，但仍然可以对其进行编程以执行大量有用的工作。

Intel’s focus had always been on semiconductor memory chips. It had shied away from logic chips like those suggested by Busicom, since it felt that markets for them were limited. Hoff’s insight was to recognize that by designing fewer logic chips with more general capabilities, one could satisfy Busicom’s needs elegantly. Hoff was inspired by the PDP-8, which had a very small set of instructions, but which its thousands of users had programmed to do a variety of things. He also recalled using an IBM 1620, a small scientific computer with an extremely limited instruction set that nevertheless could be programmed to do a lot of useful work.

霍夫提出了一种逻辑芯片，它融合了更多通用计算机的概念（图 7.3）。一个关键特性是能够调用子程序、执行子程序并根据需要返回主程序。48^他建议使用一个寄存器来实现这一点，该寄存器跟踪程序执行的位置，并在被中断执行子程序时保存该状态。子程序本身可以被中断，返回地址存储在“堆栈”中：一种寄存器排列，可自动按后进先出原则检索数据^。49

Hoff proposed a logic chip that incorporated more of the concepts of a general-purpose computer (figure 7.3). A critical feature was the ability to call up a subroutine, execute it, and return to the main program as needed.⁴⁸ He proposed to do that with a register that kept track of where a program was in its execution and saved that status when interrupted to perform a subroutine. Subroutines themselves could be interrupted, with return addresses stored on a “stack”: an arrangement of registers that automatically retrieved data on a last-in-first-out basis.⁴⁹

有了这种能力，芯片就可以执行以子程序形式存储在内存中的复杂操作，而不必将这些功能永久地连接到芯片上。按照霍夫的方法做会比较慢，但在计算器中这并不重要，因为人无论如何都无法按那么快的键。逻辑的复杂性现在将驻留在存储在内存芯片中的软件中，因此人们不会白白得到任何东西。但英特尔是一家内存公司，它知道它可以提供足够容量的内存芯片。作为额外的诱因，逻辑芯片的销售将意味着其主要内存产品的销售增加。

With this ability, the chip could carry out complex operations stored as subroutines in memory, and avoid having those functions permanently wired onto the chip. Doing it Hoff’s way would be slower, but in a calculator that did not matter, since a person could not press keys that fast anyway. The complexity of the logic would now reside in software stored in the memory chips, so one was not getting something for nothing. But Intel was a memory company, and it knew that it could provide memory chips with enough capacity. As an added inducement, sales of the logic chips would mean more sales of its bread-and-butter memories.

图 7.3

（上）英特尔公司的 ME Hoff、Stanley Mazor 和 Federico Faggin 申请的“多芯片数字计算机内存系统”专利。该专利并非专门针对“芯片上的计算机”，但请注意，该图中显示了存储程序计算机处理器中的所有功能块。（下）英特尔 8080。（来源：史密森学会。）

Figure 7.3

(top) Patent for a “Memory System for a Multi-Chip Digital Computer,” by M. E. Hoff, Stanley Mazor, and Federico Faggin of Intel. The patent was not specifically for a “computer on a chip,” but note that all the functional blocks found in the processor of a stored-program computer are shown in this drawing. (bottom) Intel 8080. (Source: Smithsonian Institution.)

这种灵活性意味着这组芯片除了用于计算器外，还可用于许多其他应用。Busicom 处于竞争激烈且动荡的市场，英特尔意识到了这一点。（Busicom 最终破产了。）罗伯特·诺伊斯与 Busicom 达成协议，以较低的成本向其提供芯片，作为回报，英特尔有权向其他客户推销这些芯片，用于非计算器应用。诺伊斯说，这些与 Busicom 的简单谈判带来了计算历史上的一个关键时刻^。50

That flexibility meant that the set of chips could be used for many other applications besides calculators. Busicom was in a highly competitive and volatile market, and Intel recognized that. (Busicom eventually went bankrupt.) Robert Noyce negotiated a deal with Busicom to provide it with chips at a lower cost, giving Intel in return the right to market the chips to other customers for noncalculator applications. From these unsophisticated negotiations with Busicom, in Noyce’s words, came a pivotal moment in the history of computing.⁵⁰

最终，他们开发出了一套四块芯片，最早在 1971 年底的一份行业杂志上刊登广告，其中包括“芯片上的微程序计算机！” ⁵¹这就是 4004，它包含了微型通用存储程序计算机的所有基本寄存器和控制功能。其他芯片包含只读存储器 (ROM)、随机存取存储器 (RAM) 和用于处理输出功能的芯片。4004 成为了历史性的里程碑，但其他芯片也同样重要，尤其是 ROM 芯片，它提供了将通用处理器变成可以满足客户需求的东西的代码。（同样在英特尔，由 Dov Frohman 领导的团队开发了一种 ROM 芯片，可以通过暴露在紫外线下轻松重新编程和擦除。这种芯片被称为 EPROM（可擦除可编程只读存储器），于 1971 年推出，它使使用微处理器进行系统设计的概念变得切实可行。）⁵²

The result was a set of four chips, first advertised in a trade journal in late 1971, which included “a microprogrammable computer on a chip!”⁵¹ That was the 4004, on which one found all the basic registers and control functions of a tiny, general-purpose stored-program computer. The other chips contained a read-only memory (ROM), random-access memory (RAM), and a chip to handle output functions. The 4004 became the historical milestone, but the other chips were important as well, especially the ROM chip that supplied the code that turned a general-purpose processor into something that could meet a customer’s needs. (Also at Intel, a team led by Dov Frohman developed a ROM chip that could be easily reprogrammed and erased by exposure to ultraviolet light. Called an EPROM (erasable programmable read-only memory) and introduced in 1971, it made the concept of system design using a microprocessor practical.)⁵²

4004 的详细设计由 Stan Mazor 完成。Federico Faggin 在将这一概念付诸实践方面也发挥了关键作用。Busicom 的代表 Masatoshi Shima 也做出了贡献。许多关于这项发明的记载都把功劳归于 Hoff；包括 Hoff 在内的所有人都同意这种说法并不准确。Faggin 于 1974 年离开英特尔，成立了一家竞争对手公司 Zilog。与 Zilog 竞争的英特尔认为没有必要在其宣传材料中宣传 Faggin 的才能，尽管英特尔从未对其前员工表现出任何敌意。53^功劳归于谁的问题揭示了许多人对发明的看法：Hoff 有将通用计算机放在芯片上的想法，而 Faggin 和其他人“仅仅”在硅片上实现了这个想法。当时，英特尔也不确定自己发明了什么：英特尔的专利律师当时拒绝了 Hoff 将该作品作为“计算机”申请专利的愿望。⁵⁴英特尔获得了 4004 的两项专利，涵盖其架构和实现；霍夫的名字只出现在其中一项上。（这为 TI 和 Gil Hyatt 争夺专利使用费打开了大门。）

The detailed design of the 4004 was done by Stan Mazor. Federico Faggin was also crucial in making the concept practical. Masatoshi Shima, a representative from Busicom, also contributed. Many histories of the invention give Hoff sole credit; all players, including Hoff, now agree that that is not accurate. Faggin left Intel in 1974 to found a rival company, Zilog. Intel, in competition with Zilog, felt no need to advertise Faggin’s talents in its promotional literature, although Intel never showed any outward hostility to its ex-employee.⁵³ The issue of whom to credit reveals the way many people think of invention: Hoff had the idea of putting a general-purpose computer on a chip, Faggin and the others “merely” implemented that idea in silicon. At the time, Intel was not sure what it had invented either: Intel’s patent attorney resisted Hoff’s desire at the time to patent the work as a “computer.”⁵⁴ Intel obtained two patents on the 4004, covering its architecture and implementation; Hoff’s name appears on only one of them. (That opened the door to rival claims for patent royalties from TI, and eventually Gil Hyatt.)

4004 每次处理 4 位数据——足以编码十进制数字，但不能再多了。在与 Busicom 合作的几乎同时，英特尔与德克萨斯州圣安东尼奥的计算机终端公司（后来称为 Datapoint）达成了类似的协议，生产一组用于连接到大型计算机的终端芯片。Mazor 和 Hoff 再次提出了一种微处理器来处理终端的逻辑。他们提出的芯片将以 8 位块的形式处理数据，足以一次处理一个完整的字节。当英特尔完成设计时，Datapoint 已决定采用传统的 TTL 芯片。英特尔于 1972 年 4 月推出了这款芯片，他们称之为 8008，作为商业产品^。55

The 4004 worked with groups of four bits at a time—enough to code decimal digits but no more. At almost the same time as the work with Busicom, Intel entered into a similar agreement with Computer Terminal Corporation (later called Datapoint) of San Antonio, Texas, to produce a set of chips for a terminal to be attached to mainframe computers. Again, Mazor and Hoff proposed a microprocessor to handle the terminal’s logic. Their proposed chip would handle data in 8-bit chunks, enough to process a full byte at a time. By the time Intel had completed its design, Datapoint had decided to go with conventional TTL chips. Intel offered the chip, which they called the 8008, as a commercial product in April 1972.⁵⁵

1972 年末，罗克韦尔推出了一款 4 位微处理器。罗克韦尔是一家汽车公司，与民兵制导系统制造商北美航空公司合并。1973 年，其他六家公司也开始提供微处理器。英特尔于 1974 年 4 月回应了竞争，宣布推出 8080，这是一款 8 位芯片，可以处理更多内存，并且比 8008 需要更少的支持芯片。该公司将价格定为 360 美元——这个数字有点随意，因为英特尔没有一次销售此类芯片的经验。（民间传说 360 美元的价格是为了与 IBM System/360 进行比较。）56 8080^比8008 有了显著的进步，可以执行为其他芯片编写的程序，这种兼容性对英特尔在市场上的主导地位至关重要。 8080 是第一款指令集和内存寻址能力接近当时小型计算机的微处理器^。57

In late 1972, a 4-bit microprocessor was offered by Rockwell, an automotive company that had merged with North American Aviation, maker of the Minuteman Guidance System. In 1973 a half dozen other companies began offering microprocessors as well. Intel responded to the competition in April 1974 by announcing the 8080, an 8-bit chip that could address much more memory and required fewer support chips than the 8008. The company set the price at $360—a somewhat arbitrary figure, as Intel had no experience selling chips like these one at a time. (Folklore has it that the $360 price was set to suggest a comparison with the IBM System/360.)⁵⁶ A significant advance over the 8008, the 8080 could execute programs written for the other chip, a compatibility that would prove crucial to Intel’s dominance of the market. The 8080 was the first of the microprocessors whose instruction set and memory addressing capability approached those of the minicomputers of the day.⁵⁷

从微处理器到个人计算机

From Microprocessor to Personal Computer

1974 年初，两股力量汇聚在一起。一方面是半导体工程师，他们拥有功能越来越强大的微处理器和容量越来越大的内存芯片。另一方面是分时系统的用户，他们将 PDP-10 或 XDS 940 视为公众使用计算的基础。当这两股力量汇聚在一起时，它们将带来个人计算的革命。

There were now, in early 1974, two converging forces at work. From one direction were the semiconductor engineers with their ever-more-powerful microprocessors and ever-more-capacious memory chips. From the other direction were users of time-sharing systems, who saw a PDP-10 or XDS 940 as a basis for public access to computing. When these forces met in the middle, they would bring about a revolution in personal computing.

它们差点没能相遇。从布兰德的观察到 Altair 的出现，这两年里，两股力量一直在较量。分时系统甚至无法从工业客户那里赚钱，而像 Community Memory 这样的公共系统也举步维艰。另一方面，半导体公司并不认为他们的产品可能成为个人电脑的基础。

They almost did not meet. For the two years between Brand’s observation and the appearance of the Altair, the two forces were rushing past one another. The time-sharing systems had trouble making money even from industrial clients, and the public systems like Community Memory were also struggling. At the other end, semiconductor companies did not think of their products as a possible basis for a personal computer.

基于微处理器的通用计算机确实在 1973 年出现。同年 5 月，从越南移民到法国的 Thi T. Truong 让他的电子公司设计并制造了基于英特尔 8008 微处理器的计算机。MICRAL 是一款坚固耐用且设计精良的计算机，具有总线架构和电路板上的内部插槽以供扩展。基本型号的价格不到 2,000 美元，它找到了一个替代小型计算机进行简单控制操作的市场。在接下来的两年里，大约售出了两千台，其中没有一台销往工业市场以外。58^它被认为是第一台在商业市场上销售的基于微处理器的计算机。由于 8008 的局限性、它在法国的位置，以及最重要的是它的创造者未能看清它“真正”是什么，它从未突破其作为有限工业地点的小型计算机替代品的利基市场。

A general-purpose computer based on a microprocessor did appear in 1973. In May of that year Thi T. Truong, an immigrant to France from Viet Nam, had his electronics company design and build a computer based on the Intel 8008 microprocessor. The MICRAL was a rugged and well-designed computer, with a bus architecture and internal slots on its circuit board for expansion. A base model cost under $2,000, and it found a market replacing minicomputers for simple control operations. Around two thousand were sold in the next two years, none of them beyond an industrial market.⁵⁸ It is regarded as the first microprocessor-based computer to be sold in the commercial marketplace. Because of the limitations of the 8008, its location in France, and above all, the failure by its creators to see what it “really” was, it never broke out of its niche as a replacement for minicomputers in limited industrial locations.

英特尔也认为 MICRAL 是 mini 的替代品。英特尔对其产品的思维模式是这样的：工业客户购买了一台 8080，并为其编写了专用软件，然后将其刻录到只读存储器中，以提供具有所需功能的系统。最终产生的廉价产品（不再可编程）随后作为工业系统中的嵌入式控制器投放市场。这种思维模式的主要原因是了解对微处理器进行编程有多难。当英特尔的传统客户硬件设计师本身不熟悉编程时，要求未经培训的消费者进行编程似乎很荒谬。

The perception of the MICRAL as something to replace the mini was echoed at Intel as well. Intel’s mental model of its product was this: an industrial customer bought an 8080 and wrote specialized software for it, which was then burned into a read-only-memory to give a system with the desired functions. The resulting inexpensive product (no longer programmable) was then put on the market as an embedded controller in an industrial system. A major reason for that mental model was the understanding of how hard it was to program a microprocessor. It seemed absurd to ask untrained consumers to program when Intel’s traditional customers, hardware designers, were themselves uncomfortable with programming.

考虑到这些嵌入式用途，微处理器供应商开发了教育套件，旨在帮助客户轻松进行系统设计。这些套件包括微处理器、一些 RAM 和 ROM 芯片，以及一些处理时序和控制的其他芯片，全部安装在印刷电路板上。它们还包括书面材料，提供有关如何编程系统的教程。这一努力使英特尔远离了其制造芯片的核心业务，但该公司希望通过批量销售组件来弥补当前的损失。59^这些套件以 200 美元左右的价格出售，或免费赠送给工程师，以便他们以后可以批量销售。

With these embedded uses in mind, microprocessor suppliers developed educational packages intended to ease customers into system design. These kits included the microprocessor, some RAM and ROM chips, and some other chips that handled timing and control, all mounted on a printed circuit board. They also included written material that gave a tutorial on how to program the system. This effort took Intel far from its core business of making chips, but the company hoped to recoup the current losses later on with volume sales of components.⁵⁹ These kits were sold for around $200 or given away to engineers who might later generate volume sales.

英特尔和其他公司还构建了更为复杂的“开发系统”，客户可以在该系统上实际测试应用程序的软件（图 7.4）。这些是完全组装好的产品，售价约为 10,000 美元。要使用这些系统，客户还需要专门的软件，以便他们使用 FORTRAN 等语言编写程序，然后为微处理器“交叉编译”它，也就是说，从 FORTRAN 程序生成机器代码，不是为编写它的计算机，而是为微处理器。该公司聘请了加利福尼亚州蒙特雷海军研究生院的讲师 Gary Kildall 来开发一种基于 IBM 的 PL/I 的语言。60^他将其称为 PL/M，英特尔于 1973 年将其提供给客户。最初，该软件旨在在大型机上运行，但很快就可以在小型计算机上使用，最后可以在基于微处理器的系统上运行。 1974 年，英特尔推出了 Intellec 4 开发系统，该系统包含自己的常驻 PL/M 编译器（即，不需要大型机或小型机来编译代码）。61^类似的 Intellec-8 推出了 8 位微处理器。

Intel and the others also built more sophisticated “Development Systems,” on which a customer could actually test the software for an application (figure 7.4). These were fully assembled products that sold for around $10,000. To use these systems, customers also needed specialized software that would allow them to write programs using a language like FORTRAN, and then “cross-compile” it for the microprocessor—that is, from the FORTRAN program generate machine code, not for the computer on which it was written, but for the microprocessor. The company hired Gary Kildall, an instructor at the Naval Postgraduate School in Monterey, California, to develop a language based on IBM’s PL/I.⁶⁰ He called it PL/M, and in 1973 Intel offered it to customers. Initially this software was intended to be run on a large mainframe, but it was soon available for minicomputers, and finally to microprocessor-based systems. In 1974 Intel offered a development system, the Intellec 4, which included its own resident PL/M compiler (i.e., one did not need a mainframe or a mini to compile the code).⁶¹ A similar Intellec-8 introduced the 8-bit microprocessors.

图 7.4

Intellec-8 开发系统。这实际上是一台通用计算机，但英特尔并没有将其作为通用计算机进行营销。英特尔希望客户购买它们来帮助编写和调试将用于嵌入式系统的微处理器软件。一些客户购买并用作小型计算机的替代品。（来源：英特尔。）

Figure 7.4

Intellec-8 Development System. This was, in fact, a general-purpose computer, but Intel did not market it as such. Intel intended that customers buy them to assist in writing and debugging microprocessor software that would go into embedded systems. A few were purchased and used as alternatives to minicomputers. (Source: Intel.)

通过这些开发系统，英特尔实际上发明了个人计算机。但该公司并没有意识到这一点。这些套件在销售时并没有作为功能齐全的计算机。偶尔有人会购买其中一种系统并用它代替小型计算机，但英特尔既不支持这种努力，也不承认它的潜力。62^英特尔和其他微处理器公司通过销售这些开发系统赚钱——对某些公司来说，它们非常有利可图——但他们的目标是利用它们作为杠杆，打开芯片的批量购买渠道。公众是买不到的。芯片供应商关注的是让嵌入式系统完成有用工作的困难；他们认为，公众不会愿意为了拥有自己的计算机而忍受编程的困难。

With these development systems, Intel had in fact invented a personal computer. But the company did not realize it. These kits were not marketed as the functional computers they were. Occasionally someone bought one of these systems and used it in place of a minicomputer, but Intel neither supported that effort nor recognized its potential.⁶² Intel and the other microprocessor firms made money selling these development systems—for some they were very profitable—but the goal was to use them as a lever to open up volume purchases of chips. The public could not buy one. The chip suppliers were focused on the difficulties in getting embedded systems to do useful work; they did not think that the public would be willing to put up with the difficulties of programming just to own their own computer.

爱好者的作用

Role of Hobbyists

这时电子爱好者和发烧友就派上用场了。如果没有他们，个人计算领域的两股力量可能就会交叉，而不会汇聚在一起。那时，爱好者们愿意做让基于微处理器的系统变得实用所需的工作。

Here is where the electronics hobbyists and enthusiasts come in. Were it not for them, the two forces in personal computing might have crossed without converging. Hobbyists, at that moment, were willing to do the work needed to make microprocessor-based systems practical.

这个群体有着悠久的技术创新历史——例如，第一次世界大战后，无线电爱好者开辟了高频无线电频谱，用于长距离无线电通信。第二次世界大战后，业余无线电爱好的范围从业余无线电扩展到高保真音乐重放、自动控制和简单的机器人。美国陆军通信兵团的大量战争剩余设备流入个人手中，进一步助长了这一现象。（曼哈顿下城的一个街区被称为“无线电街”，世贸中心就建在这里，是剩余电子设备的著名来源。）⁶³从真空管到集成电路的转变使得个人更难在家中用面包板搭建电路，但廉价的 TTL 芯片现在本身就包含整个电路。64^随着业余爱好从模拟应用迅速发展到数字应用，这个群体为个人电脑的创造提供了一个关键组成部分：它提供了计算机公司和芯片制造商都无法提供的支持基础设施。

This community had a long history of technical innovation—it was radio amateurs, for example, who opened up the high-frequency radio spectrum for long-distance radio communications after World War I. After World War II, the hobby expanded beyond amateur radio to include high-fidelity music reproduction, automatic controls, and simple robotics. A cornucopia of war surplus equipment from the U.S. Army Signal Corps found its way into individual hands, further fueling the phenomenon. (A block in lower Manhattan known as “Radio Row,” where the World Trade Center was built, was a famous source of surplus electronic gear.)⁶³ The shift from vacuum tubes to integrated circuits made it harder for an individual to build a circuit on a breadboard at home, but inexpensive TTL chips now contained whole circuits themselves.⁶⁴ As the hobby evolved rapidly from analog to digital applications, this group supplied a key component in creating the personal computer: it provided an infrastructure of support that neither the computer companies nor the chip makers could.

这种基础设施包括各种电子杂志。有些杂志针对特定群体，例如针对无线电爱好者的 QS T。其中两本，《大众电子》和《无线电电子》，是大众感兴趣的，在报摊上出售；它们涵盖高保真音频、短波广播、电视以及各种家用和车载小工具。每期通常至少有一个建设项目。对于这些项目，该杂志将与小型电子公司安排提供已经蚀刻和钻孔的印刷电路板，以及读者可能难以在当地找到的专用组件。通过扫描这些杂志的往期杂志，我们可以追踪业余爱好者如何从模拟设计转向数字设计。

This infrastructure included a variety of electronics magazines. Some were aimed at particular segments, for example, QST for radio amateurs. Two of them, Popular Electronics and Radio-Electronics, were of general interest and sold at newsstands; they covered high-fidelity audio, shortwave radio, television, and assorted gadgets for the home and car. Each issue typically had at least one construction project. For these projects the magazine would make arrangements with small electronics companies to supply a printed circuit board, already etched and drilled, as well as specialized components that readers might have difficulty finding locally. By scanning the back issues of these magazines we can trace how hobbyists moved from analog to digital designs.

1971 年 9 月的《科学美国人》杂志刊登了一款名为 Kenbak-1 的机器的广告，该机器由中型和小型集成电路制成。广告称它适合“个人”，但实际上它是为学校设计的。Kenbak 可能是第一台个人电脑，但它没有使用微处理器，功能也相当有限。

A machine called the Kenbak-1, made of medium and small-scale integrated circuits, was advertised in the September 1971 issue of Scientific American. The advertisement called it suitable for “private individuals,” but it was really intended for schools. The Kenbak may be the first personal computer, but it did not use a microprocessor, and its capabilities were quite limited.

Scelbi-8H 是在 1974 年 3 月QST杂志背面的一则小广告中发布的。它使用了英特尔 8008，因此可能是第一台面向公众销售的基于微处理器的计算机。根据广告，“新款 Scelbi-8H 微型计算机的套件价格低至 440 美元！” ⁶⁵目前尚不清楚 Scelbi 销售了多少台机器，但该公司在早期个人电脑现象中发挥了重要作用^。66

The Scelbi-8H was announced in a tiny advertisement in the back of the March 1974 issue of QST. It used an Intel 8008, and thus may be the first microprocessor-based computer marketed to the public. According to the advertisement, “Kit prices for the new Scelbi-8H mini-computer start as low as $440!”⁶⁵ It is not known how many machines Scelbi sold, but the company went on to play an important part in the early personal computer phenomenon.⁶⁶

1974 年 7 月，《无线电电子》杂志以“构建 Mark-8：您的个人小型计算机”为标题，发布了一款基于英特尔 8008 的套件。67^该项目比该杂志通常刊登的内容要雄心勃勃得多。文章只作了简单的描述，并要求读者订购一本单独的 5.00 美元的小册子以获取详细说明。Mark-8 由布莱克斯堡弗吉尼亚理工大学的 Jonathan Titus 设计。实际制造的机器数量可能在数百台左右，尽管据报道该杂志售出了“数千”本小册子。丹佛至少出现了一个 Mark-8 用户俱乐部，其成员设计了一种在录音机上存储程序的巧妙方法。68^读者被引导到新泽西州恩格尔伍德的一家公司，该公司以 47.50 美元的价格提供一套电路板，并向英特尔购买 8008 芯片（售价 120.00 美元）。 Mark-8 在《无线电电子学》杂志上的亮相是其竞争对手《大众电子学》在六个月后决定推出 Altair 套件的一个重要因素^。69

In July 1974, Radio-Electronics announced a kit based on the Intel 8008, under the headline “Build the Mark-8: Your Personal Minicomputer.”⁶⁷ The project was much more ambitious than what typically appeared in that magazine. The article gave only a simple description and asked readers to order a separate, $5.00 booklet for detailed instructions. The Mark-8 was designed by Jonathan Titus of Virginia Polytechnic University in Blacksburg. The number of machines actually built may range in the hundreds, although the magazine reportedly sold “thousands” of booklets. At least one Mark-8 users club sprang up, in Denver, whose members designed an ingenious method of storing programs on an audio cassette recorder.⁶⁸ Readers were directed to a company in Englewood, New Jersey, that supplied a set of circuit boards for $47.50, and to Intel for the 8008 chip (for $120.00). The Mark-8’s appearance in Radio-Electronics was a strong factor in the decision by its rival Popular Electronics to introduce the Altair kit six months later.⁶⁹

这些套件只是爱好者杂志中介绍的众多项目中的一小部分。它们反映了社区有意识的努力，将数字电子产品，以及它的所有前景和复杂性，带给只熟悉简单的无线电或音频设备的业余爱好者。这不是一个容易的转变：Mark-8 和电视打字机（下文将描述）的构造过于复杂，无法在一篇杂志文章中描述；读者必须订购单独的小册子才能获得完整的计划。《无线电电子》向读者解释说：“我们不打算将这种方式作为常规做法来写文章。” ⁷⁰尽管数字电路比该杂志处理的要复杂得多，但它认识到电子世界正在朝这个方向发展，读者想要这样的项目。

These kits were just a few of many projects described in the hobbyist magazines. They reflected a conscious effort by the community to bring digital electronics, with all its promise and complexity, to amateurs who were familiar only with simpler radio or audio equipment. It was not an easy transition: construction of both the Mark-8 and the TV-typewriter (described next) was too complex to be described in a magazine article; readers had to order a separate booklet to get complete plans. Radio-Electronics explained to its readers that “[w]e do not intend to do an article this way as a regular practice.”⁷⁰ Although digital circuits were more complex than what the magazine had been handling, it recognized that the electronics world was moving in that direction and that its readers wanted such projects.

其他文章描述了使用廉价 TTL 芯片的更简单的数字设备——计时器、游戏、时钟、键盘和测量仪器。一个有影响力的项目是电视打字机，由唐·兰开斯特设计，并于 1973 年 9 月发表在《无线电电子学》上。该设备允许读者在普通电视机上显示以 ASCII 编码的字母数字字符。它预示着 CRT 终端作为个人电脑的主要输入输出设备的出现——这是 PC 文化与依赖电传打字机的小型计算机文化之间的一个主要区别。李·费尔森斯坦称电视打字机为“计算机革命的开端” ^。71

Other articles described simpler digital devices—timers, games, clocks, keyboards, and measuring instruments—that used inexpensive TTL chips. One influential project was the TV-Typewriter, designed by Don Lancaster and published in Radio-Electronics in September 1973. This device allowed readers to display alphanumeric characters, encoded in ASCII, on an ordinary television set. It presaged the advent of CRT terminals as the primary input-output device for personal computers—one major distinction between the PC culture and that of the minicomputer, which relied on the Teletype. Lee Felsenstein called the TV-Typewriter “the opening shot of the computer revolution.”⁷¹

牵牛星

Altair

1974 年是个人计算的奇迹之年。1 月，惠普推出了 HP-65 可编程计算器。同年夏天，英特尔发布了 8080 微处理器。7 月，《无线电电子》杂志介绍了 Mark-8。12 月下旬，《大众电子》的订阅者通过邮件收到了 1975 年 1 月刊，封面上是“Altair”小型计算机的原型（图 7.5），还有一篇文章介绍了读者如何以不到 400 美元的价格获得一台这样的计算机。这一发布与 IBM 十年前发布的 System/360 并列为计算历史上最重要的发布之一。但十年的时间带来了多大的变化：Altair 是一台真正的个人计算机。

1974 was the annus mirabilis of personal computing. In January, Hewlett-Packard introduced its HP-65 programmable calculator. That summer Intel announced the 8080 microprocessor. In July, Radio-Electronics described the Mark-8. In late December, subscribers to Popular Electronics received their January 1975 issue in the mail, with a prototype of the “Altair” minicomputer on the cover (figure 7.5), and an article describing how readers could obtain one for less than $400. This announcement ranks with IBM’s announcement of the System/360 a decade earlier as one of the most significant in the history of computing. But what a difference a decade made: the Altair was a genuine personal computer.

Altair 的设计者 H. Edward Roberts 理应被誉为个人电脑的发明者。Altair 是一款功能强大、价格低廉的电脑，采用 Intel 8080 微处理器设计而成。尽管将 Roberts 称为发明者只有在考虑到他之前的所有发明（包括上述关键步骤）的情况下才有意义，但他确实值得称赞。马克·吐温 (Mark Twain) 曾说，历史学家必须重新安排过去的事件，使其更有意义。如果真是这样，那么在阿尔伯克基的一家小型模型火箭爱好商店发明的个人电脑就迫切需要一些创造性的重新安排。它的完全不可能性和不可预测性导致一些人将这项发明归功于许多其他地方，这些地方更为合理，例如施乐帕洛阿尔托研究中心、数字设备公司，甚至 IBM。但阿尔伯克基是个人电脑的发明者，因为只有在 MITS，个人电脑的技术和社会组成部分才得以融合。

H. Edward Roberts, the Altair’s designer, deserves credit as the inventor of the personal computer. The Altair was a capable, inexpensive computer designed around the Intel 8080 microprocessor. Although calling Roberts the inventor makes sense only in the context of all that came before him, including the crucial steps described above, he does deserve the credit. Mark Twain said that historians have to rearrange past events so they make more sense. If so, the invention of the personal computer at a small model-rocket hobby shop in Albuquerque cries out for some creative rearrangement. Its utter improbability and unpredictability have led some to credit many other places with the invention, places that are more sensible, such as the Xerox Palo Alto Research Center, or Digital Equipment Corporation, or even IBM. But Albuquerque it was, for it was only at MITS that the technical and social components of personal computing converged.

图 7.5

MITS Altair 8800 计算机。前面板是从 Data General Nova 复制而来。这张照片中显示的机器是最早生产的机器之一，所有者是 Forrest Mims，他是一位电子爱好者，经常为《大众电子》杂志撰稿，曾在 MITS 短暂工作过。（来源：史密森学会。）

Figure 7.5

MITS Altair 8800 Computer. The front panel was copied from the Data General Nova. The machine shown in this photograph was one of the first produced and was owned by Forrest Mims, an electronics hobbyist and frequent contributor to Popular Electronics, who had briefly worked at MITS. (Source: Smithsonian Institution.)

首先考虑技术方面。其他业余爱好者项目都没有 Altair 的发布产生影响。为什么？原因之一是它被设计和宣传为功能强大的小型计算机，与 DEC 或 Data General 提供的计算机一样强大。由 Ed Roberts 和 William Yates 撰写的杂志文章反复强调了这一点：“一台可以与复杂的小型计算机相媲美的成熟计算机”；“不是‘演示器’或增强型计算器”；“性能可与当前的商用小型计算机相媲美。” ⁷² Altair 计算机的外观表明了它的小型计算机血统。它看起来像 Data General Nova：它有一个矩形金属外壳，前面板上有控制内部寄存器内容的开关，还有指示二进制 1 或 0 存在的小灯。在 Altair 的机壳内，有一台主要由 TTL 集成电路构成的机器（微处理器除外，它是 MOS 器件），采用双列直插式封装，焊接在电路板上。信号和电力通过总线从机器的一个部分传输到另一个部分。Altair 使用集成电路而不是磁芯作为主存储器。《大众电子》封面称 Altair 是“世界上第一台小型计算机套件”；除了它使用微处理器之外，它准确地描述了它的物理构造和设计^。73

Consider first the technical. None of the other hobbyist projects had the impact of the Altair’s announcement. Why? One reason was that it was designed and promoted as a capable minicomputer, as powerful as those offered by DEC or Data General. The magazine article, written by Ed Roberts and William Yates, makes this point over and over: “a full-blown computer that can hold its own against sophisticated minicomputers”; “not a ‘demonstrator’ or a souped-up calculator”; “performance competes with current commercial minicomputers.”⁷² The physical appearance of the Altair computer suggested its minicomputer lineage. It looked like the Data General Nova: it had a rectangular metal case, a front panel of switches that controlled the contents of internal registers, and small lights indicating the presence of a binary one or zero. Inside the Altair’s case, there was a machine built mainly of TTL integrated circuits (except for the microprocessor, which was a MOS device), packaged in dual-in-line packages, soldered onto circuit boards. Signals and power traveled from one part of the machine to another on a bus. The Altair used integrated circuits, not magnetic cores, for its primary memory. The Popular Electronics cover called the Altair the “world’s first minicomputer kit”; except for its use of a microprocessor, that accurately described its physical construction and design.⁷³

但广告中宣传的 Altair 比 1975 年的小型计算机便宜 10 倍。杂志以不到 400 美元的价格出售了 Altair 套件，还有几百台已组装好的。杂志封面上说读者可以“节省 1,000 多美元”。事实上，最便宜的 PDP-8 也要几千美元。当然，PDP-8 是一台完全组装好的、可操作的计算机，其功能比基本型 Altair 强大得多，但在这种情况下这并不重要。（400 美元到底能买到什么，后面会讨论。）低成本主要源于它使用了刚刚推出的 Intel 8080 微处理器。Intel 曾对小批量 8080 给出 360 美元的报价，但 Intel 的报价并非基于对如何向这个市场销售 8080 的仔细分析。MITS 只以每台 75 美元的价格购买了它们^。74

But the Altair as advertised was ten times cheaper than minicomputers were in 1975. The magazine offered an Altair for under $400 as a kit, and a few hundred more already assembled. The magazine cover said that readers could “save over $1,000.” In fact, the cheapest PDP-8 cost several thousand dollars. Of course, a PDP-8 was a fully assembled, operating computer that was considerably more capable than the basic Altair, but that did not really matter in this case. (Just what one got for $400 will be discussed later.) The low cost resulted mainly from its use of the Intel 8080 microprocessor, just introduced. Intel had quoted a price of $360 for small quantities of 8080s, but Intel’s quote was not based on a careful analysis of how to sell the 8080 to this market. MITS bought them for only $75 each.⁷⁴

8080 具有更多指令，速度更快，性能比 Mark-8 和 Scelbi-8 使用的 8008 更强。它还允许更简单的设计，因为它只需要六个而不是二十个支持芯片就可以组成一个功能系统。8008 的其他改进包括它能够寻址高达 64,000 字节的内存（而 8008 为 16,000 字节），并且它使用主内存作为堆栈，这允许无限级别的子程序，而不是 8008 的七个级别。

The 8080 had more instructions and was faster and more capable than the 8008 that the Mark-8 and Scelbi-8 used. It also permitted a simpler design since it required only six instead of twenty supporting chips to make a functional system. Other improvements over the 8008 were its ability to address up to 64 thousand bytes of memory (vs. the 8008’s 16 thousand), and its use of main memory for the stack, which permitted essentially unlimited levels of subroutines instead of the 8008’s seven levels.

8080 处理器只是 Altair 相对于其前辈的架构优势之一。同样重要的是它使用了开放总线。75^据传言，总线架构几乎没有出现。在制造出 Altair 原型机后，罗伯茨对其进行了拍照，并通过 Railway Express 将其运送到纽约的 Popula r Electronics办公室。Railway Express 是美国早期工业革命的遗迹，当时它即将破产；它丢失了这批货物。杂志封面展示了原型机，其浅色前面板和左上角的“Altair 8800”字样。这台机器有一组四块大电路板，一块叠一块，一条宽带状电缆将 100 条线从一块电路板传输到另一块电路板。在那台机器丢失后，罗伯特重新设计了 Altair。他换用了更大的深蓝色机柜，并丢弃了 100 线带状电缆。在新设计中，连接到刚性背板的电线将信号从一块电路板传输到另一块电路板。这样一来，爱好者们就可以添加一组连接器，以便接受除最初四个卡之外的其他卡^。76

The 8080 processor was only one architectural advantage the Altair had over its predecessors. Just as important was its use of an open bus.⁷⁵ According to folklore, the bus architecture almost did not happen. After building the prototype Altair, Roberts photographed it and shipped it via Railway Express to the offices of Popular Electronics in New York. Railway Express, a vestige of an earlier American industrial revolution, was about to go bankrupt; it lost the package. The magazine cover issue showed the prototype, with its light-colored front panel and the words “Altair 8800” on the upper left. That machine had a set of four large circuit boards stacked on top of one another, with a wide ribbon cable carrying 100 lines from one board to another. After that machine was lost, Robert redesigned the Altair. He switched to a larger deep blue cabinet and discarded the 100-wire ribbon cable. In the new design, wires connected to a rigid backplane carried the signals from one board to another. That allowed hobbyists to add a set of connectors that could accept other cards besides the initial four.⁷⁶

售价 400 美元的套件只附带两块可插入总线的卡：这两块卡加上一块用于控制前面板和电源的电路板，就构成了整台计算机。内部看上去很空。但费力地将一组电线焊接到扩展底盘上，就形成了一整套插槽，可以插入很多卡。MITS 当时已经在设计用于更多内存、I/O 和其他功能的卡。

The $400 kit came with only two cards to plug into the bus: those two, plus a circuit board to control the front panel and the power supply, made up the whole computer. The inside looked quite bare. But laboriously soldering a set of wires to an expansion chassis created a full set of slots into which a lot of cards could be plugged. MITS was already designing cards for more memory, I/O and other functions.

按照数字设备公司建立的传统，罗伯茨没有将总线规格作为公司机密。这样其他人就可以设计和销售 Altair 的卡。这一决定对 Altair 的成功至关重要，就像它选择 8080 处理器一样。这也解释了 Altair 的一大讽刺之处：它开创了 PC 时代，尽管它既不可靠，设计也不太好。如果其他公司不能提供改进 MITS 原始设计的插卡，Altair 的影响可能不会比 Mark-8 更大。总线架构也导致了该公司几年后的倒闭，因为它允许其他公司销售兼容卡，后来还销售兼容计算机。但那时闸门已经打开。如果 MITS 无法兑现将 Altair 打造成一台严肃机器的承诺（尽管它努力了），其他公司就会介入。MITS 继续开发插卡和外围设备，但大量订单对这家小公司来说实在是太多了。

Following the tradition established by Digital Equipment Corporation, Roberts did not hold specifications of the bus as a company secret. That allowed others to design and market cards for the Altair. That decision was as important to the Altair’s success as its choice of an 8080 processor. It also explains one of the great ironies of the Altair, that it inaugurated the PC era although it was neither reliable nor very well-designed. Had it not been possible for other companies to offer plug-in cards that improved on the original MITS design, the Altair might have made no greater impact than the Mark-8 had. The bus architecture also led to the company’s demise a few years later, since it allowed other companies to market compatible cards and, later, compatible computers. But by then the floodgates had opened. If MITS was unable to deliver on its promises of making the Altair a serious machine (though it tried), other companies would step in. MITS continued developing plug-in cards and peripheral equipment, but the flood of orders was too much for the small company.

因此，虽然爱好者花 400 美元买不到什么，但他们可以买下剩下的东西，或者自己设计和组装剩下的东西。将计算机作为基本套件进行营销，为成千上万的人提供了一种进入计算机时代的方法，而速度是他们自己（而不是计算机公司）可以控制的。

So while it was true that for $400 hobbyists got very little, they could get the rest—or design and build the rest. Marketing the computer as a bare-bones kit offered a way for thousands of people to bootstrap their way into the computer age, at a pace that they, not a computer company, could control.

组装 Altair 比组装其他电子套件（如 Heath Company 或 Dynaco 出售的套件）困难得多。MITS 以 498 美元的价格出售“完全组装和测试”的计算机，但由于订单积压如此之多，读者面临着选择：是订购套件并在几个月内收到货，还是订购组装好的计算机并可能等待一年或更长时间。77^大多数人订购了套件，并寻求彼此的帮助，以发现不可避免的接线错误和焊接不良的连接。该杂志文章针对的电子爱好者读者群，他们不是将 Altair 与简单的 Heathkits 进行比较，而是将其与从头开始构建计算机进行比较，这几乎是不可能的：不仅设计计算机很困难，而且不可能获得必要的芯片。芯片价格低廉，但前提是大量购买，而且大多数半导体公司没有为单件或零售建立分销渠道。部分原因是，客户正确地感觉到他们得到了一笔难以置信的便宜货。

Assembling the Altair was much more difficult than assembling other electronics kits, such as those sold by the Heath Company or Dynaco. MITS offered to sell “completely assembled and tested” computers for $498, but with such a backlog of orders, readers were faced with the choice of ordering the kit and getting something in a couple of months, or ordering the assembled computer and perhaps waiting a year or more. ⁷⁷ Most ordered the kit and looked to one another for support in finding the inevitable wiring errors and poorly soldered connections that they would make. The audience of electronics hobbyists, at whom the magazine article was aimed, compared the Altair not to the simple Heathkits, but to building a computer from scratch, which was almost impossible: not only was it hard to design a computer, it was impossible to obtain the necessary chips. Chips were inexpensive, but only if they were purchased in large quantities, and anyway, most semiconductor firms had no distribution channels set up for single unit or retail sales. Partly because of this, customers felt, rightly, that they were getting an incredible bargain.

基本型 Altair 功能有限，再加上《大众电子学》文章发表时唯一一台 Altair 已失传，人们认为它是假货，是“骗子”，根本不是正经产品。78 ^Altair的创造者本意是要制造一台功能与当时市场上的微型计算机相当的正经计算机。事实证明，制造这些计算机比他们预想的要困难得多。幸运的是，业余爱好者明白这一点。但毫无疑问：Altair 是真实存在的。

The limited capabilities of the basic Altair, plus the loss of the only existing Altair by the time the Popular Electronics article appeared, led to the notion that it was a sham, a “humbug,” not a serious product at all.⁷⁸ The creators of the Altair fully intended to deliver a serious computer whose capabilities were on a par with minicomputers then on the market. Making those deliveries proved to be a lot harder than they anticipated. Fortunately, hobbyists understood that. But there should be no mistake about it: the Altair was real.

MITS 公司和《大众电子》的编辑们找到了一种将集成电路的惊人进步带给个人的方法。第一批客户是业余爱好者，他们用这些机器做的第一件事就是玩游戏。79^然而，罗伯茨试图把它作为一台用于严肃工作的机器来销售。在《大众电子》的文章中，他提出了 23 种应用程序，其中没有一种是游戏。80^因为几年后才有人能提供外围设备、内存和软件，所以严肃的应用程序最初很少见。再加上 Mark-8 等其他机器的原始功能，再次导致人们认为 Altair 不是一台严肃的计算机。1975 年文章中提到的许多拟议应用程序最终都得到了实施。多年后，人们仍然可以偶尔发现 Altair（或更常见的是 Altair 克隆版）嵌入到系统中，就像它的小型计算机表亲一样。

MITS and the editors of Popular Electronics had found a way to bring the dramatic advances in integrated circuits to individuals. The first customers were hobbyists, and the first thing they did with these machines, once they got them running, was play games.⁷⁹ Roberts was trying to sell it as a machine for serious work, however. In the Popular Electronics article he proposed a list of twenty-three applications, none of them games.⁸⁰ Because it was several years before anyone could supply peripheral equipment, memory, and software, serious applications were rare at first. That, combined with the primitive capabilities of other machines like the Mark-8, led again to an assumption that the Altair was not a serious computer. Many of the proposed applications hinted at in the 1975 article were eventually implemented. Years later one could still find an occasional Altair (or more frequently, an Altair clone) embedded into a system just like its minicomputer cousins.

接下来的三年，从 1975 年 1 月到 1977 年底，计算机领域迎来了前所未有的活力和创造力。Altair 打开了闸门，尽管它的缺点人尽皆知。人们所能做的只是让它在前面板上闪烁一个灯光图案。而且这还不容易：人们必须为每个程序步骤拨动拨动开关，然后将该数字存入内存位置，然后重复该操作以执行下一步，依此类推——希望在执行过程中电源不会断电——直到整个程序（不到 256 字节长！）都在内存中。拨动小拨动开关时弄伤手指是最不令人沮丧的事情。尽管如此，总线架构意味着其他公司可以设计电路板来弥补这些缺点，甚至可以像 IMSAI 和其他公司一样设计 Altair 本身的副本^。81

The next three years, from January 1975 through the end of 1977, saw a burst of energy and creativity in computing that had almost no equal in its history. The Altair had opened the floodgates, even though its shortcomings were clear to everyone. One could do little more than get it to blink a pattern of lights on the front panel. And even that was not easy: one had to flick the toggle switches for each program step, then deposit that number into a memory location, then repeat that for the next step, and so on—hopefully the power did not go off while this was going on—until the whole program (less than 256 bytes long!) was in memory. Bruised fingers from flipping the small toggle switches were the least of the frustrations. In spite of all that, the bus architecture meant that other companies could design boards to remedy each of these shortcomings, or even design a copy of the Altair itself, as IMSAI and others did.⁸¹

但 MITS 的员工和他们的追随者们创造的不仅仅是一台计算机。这台售价 400 美元的计算机激发了用户团体、非正式通讯、商业杂志、当地俱乐部、会议甚至零售店的广泛支持。这种社交活动远远超出了传统的计算机用户团体，如 IBM 的 SHARE 或 Digital 的 DECUS。与计算器用户团体一样，这些团体也是开放和非正式的，为新手提供了更多服务。所有这些都是随着 Altair 而兴起的，许多出版物和团体在最后一台 Altair 计算机售出后还存在了很长时间。

But the people at MITS and their hangers-on created more than just a computer. This $400 computer inspired the extensive support of user groups, informal newsletters, commercial magazines, local clubs, conventions, and even retail stores. This social activity went far beyond traditional computer user groups, like SHARE for IBM or DECUS for Digital. Like the calculator users groups, these were open and informal, and offered more to the neophyte. All of this sprang up with the Altair, and many of the publications and groups lived long after the last Altair computer itself was sold.

其他公司，从 Processor Technology 开始，很快开始提供插件板，为机器提供更多内存。另一块电路板提供了一种将机器连接到电传打字机的方法，这样手指就可以恢复了。但电传打字机并不容易买到——不属于公司或大学的个人只能买到二手的，而且价格昂贵。不久之后，业余爱好者领导的小公司开始提供连接电视机和键盘的方法（尽管这些公司遵循的不是 Don Lancaster 的电视打字机的设计）。连接到电传打字机的电路板以串行方式发送数据——一次一位；另一块电路板设计为并行发送数据，用于连接到小型计算机使用的行式打印机，尽管像电传打字机一样，这些打印机价格昂贵且难以买到^。82

Other companies, beginning with Processor Technology, soon began offering plug-in boards that gave the machine more memory. Another board provided a way of connecting the machine to a Teletype, which allowed fingers to heal. But Teletypes were not easy to come by—an individual not affiliated with a corporation or university could only buy one secondhand, and even then they were expensive. Before long, hobbyists-led small companies began offering ways of hooking up a television set and a keyboard (although Don Lancaster’s TV Typewriter was not the design these followed). The board that connected to the Teletype sent data serially—one bit at a time; another board was designed that sent out data in parallel, for connection to a line printer that minicomputers used, although like the Teletype these were expensive and hard to come by.⁸²

当电源关闭时，Altair 会丢失数据，但不久之后，MITS 公司设计出一种接口，可以将数据以音调的形式输出，以便将程序存储在廉价的录音带上。1975 年底，一群业余爱好者在堪萨斯城开会，为存储在录音带上的音调制定了“堪萨斯城标准”，这样就可以在计算机之间交换程序。83^一些公司推出了无需购买电传打字机的廉价纸带阅读器。其他公司则开发了类似旧式 8 轨录音系统（将磁带一圈一圈地循环播放）的磁带盒。磁带存储速度慢且麻烦 — 用户通常必须录制多个程序副本并尝试多次才能成功将其载入计算机。在“软盘”出现之前，大容量存储不足限制了个人电脑的普及。

The Altair lost its data when the power was shut off, but before long MITS designed an interface that put out data as audio tones, to store programs on cheap audio cassettes. A group of hobbyists met in Kansas City in late 1975 and established a “Kansas City Standard” for the audio tones stored on cassettes, so that programs could be exchanged from one computer to another.⁸³ Some companies brought out inexpensive paper tape readers that did not require the purchase of a Teletype. Others developed a tape cartridge like the old 8-track audio systems, which looped a piece of tape around and around. Cassette storage was slow and cumbersome—users usually had to record several copies of a program and make several tries before successfully loading it into the computer. Inadequate mass storage limited the spread of PCs until the “floppy” disk was adapted.

软盘是由 IBM 的 David L. Noble 发明的，其用途完全不同。当 IBM 推出使用半导体存储器的 System/370 时，它需要一种方法来存储计算机的初始控制程序以及保存机器的微程序。对于使用磁芯的 System/360 来说，这不是问题，磁芯可以在断电时保存其内容。出于这种需要，IBM 于 1971 年发布了直径为 8 英寸的软盘。84^{不久之后，人们意识到软盘除了其发明时有限的用途外，还可以用于其他用途}^。特别是，曾在 IBM 工作过的 Alan Shugart 认识到软盘的简单性和低成本使其成为低成本计算机系统的理想存储介质。85尽管如此，在个人计算机发展的最初几年里，软盘驱动器还是很少见的。IBM 的硬件创新还不够；系统软件方面必须有同等的创新才能使软盘实用。在讲述这个故事之前，我们首先应该看一下更紧迫的问题，即为 PC 开发一种高级语言。

The floppy was invented by David L. Noble at IBM for a completely different purpose. When IBM introduced the System/370, which used semiconductor memory, it needed a way to store the computer’s initial control program, as well as to hold the machine’s microprogram. That had not been a problem for the System/360, which used magnetic cores that held their contents when the power was switched off. From this need came the 8-inch diameter flexible diskette, which IBM announced in 1971.⁸⁴ Before long, people recognized that it could be used for other purposes besides the somewhat limited one for which it had been invented. In particular, Alan Shugart, who had once worked for IBM, recognized that the floppy’s simplicity and low cost made it the ideal storage medium for low-cost computer systems.⁸⁵ Nevertheless, floppy drives were rare in the first few years of personal computing. IBM’s hardware innovation was not enough; there had to be an equivalent innovation in system software to make the floppy practical. Before that story is told, we shall look first at the more immediate issue of developing a high-level language for the PC.

软件：BASIC

Software: BASIC

缺乏实用的大容量存储设备是阻碍个人交互式计算普及的两大障碍之一。另一个障碍是编写应用软件的方法。到 1977 年，两款卓越且影响深远的软件——Microsoft BASIC 和 CP/M 操作系统——克服了这些障碍。

The lack of a practical mass storage device was one of two barriers that blocked the spread of personal, interactive computing. The other was a way to write applications software. By 1977 two remarkable and influential pieces of software—Microsoft BASIC and the CP/M Operating System—overcame those barriers.

在创建 Altair 时，Ed Roberts 必须做出许多选择：使用哪种处理器、总线设计（甚至是否使用总线）、封装等等。其中一个决定是选择编程语言。鉴于 BASIC 的广泛接受，很难想象曾经有过选择，但事实确实如此。BASIC 不是为小型计算机发明的。BASIC 的创建者痛恨其他人为了将该语言强行塞入比大型机更小的系统而做出的改变。即使在大型机版本中，BASIC 也存在严重的限制 - 例如，它允许的变量数量和类型。在学术计算机科学家看来，为小型计算机开发的 BASIC 版本甚至更糟 - 充满了临时补丁和修改。许多教授贬低 BASIC 是一种玩具语言，会培养不良的编程习惯，他们拒绝教授它。严肃的编程是用 FORTRAN 完成的 - 一种古老而受人尊敬但仍然功能强大的语言。

In creating the Altair, Ed Roberts had to make a number of choices: what processor to use, the design of the bus (even whether to use a bus at all), the packaging, and so on. One such decision was the choice of a programming language. Given the wide acceptance of BASIC it is hard to imagine that there ever was a choice, but there was. BASIC was not invented for small computers. The creators of BASIC abhorred the changes others made to shoehorn the language onto systems smaller than a mainframe. Even in its mainframe version, BASIC had severe limitations—on the numbers and types of variables it allowed, for example. In the view of academic computer scientists, the versions of BASIC developed for minicomputers were even worse—full of ad hoc patches and modifications. Many professors disparaged BASIC as a toy language that fostered poor programming habits, and they refused to teach it. Serious programming was done in FORTRAN—an old and venerable but still capable language.

如果有人在 1974 年要求用一种现代、简洁、设计良好的语言来取代 FORTRAN，答案可能是 APL，这是 Kenneth Iverson 在 20 世纪 60 年代初在 IBM 发明的一种交互式语言。IBM 的一个团队在 1973 年设计了一款支持 APL 的个人电脑，即“SCAMP”，尽管这款电脑的商业版本销量不佳。86^或PL/I：IBM 已将其资源投入到这种语言中，希望它能取代 FORTRAN 和 COBOL。Gary Kildall 为英特尔微处理器开发套件选择了 PL/I 的一个子集。

If, in 1974, one asked for a modern, concise, well-designed language to replace FORTRAN, the answer might have been APL, an interactive language invented at IBM by Kenneth Iverson in the early 1960s. A team within IBM designed a personal computer in 1973 that supported APL, the “SCAMP,” although a commercial version of that computer sold poorly.⁸⁶ Or PL/I: IBM had thrown its resources into this language, which it hoped would replace both FORTRAN and COBOL. Gary Kildall chose a subset of PL/I for the Intel microprocessor development kit.

BASIC 的优势在于它易于学习。更重要的是，它已经在内存有限的计算机上运行过。罗伯茨表示，在决定 Altair 搭载 BASIC 之前，他曾考虑过 FORTRAN 和^APL。87

BASIC’s strength was that it was easy to learn. More significant, it already had a track record of running on computers with limited memory. Roberts stated that he had considered FORTRAN and APL, before he decided the Altair was to have BASIC.⁸⁷

威廉·盖茨三世出生于 1955 年，当时 FORTRAN 的工作才刚刚开始。当时他还是一名哈佛学生，《大众电子学》杂志刊登了一篇介绍 Altair 的著名封面。据一位传记作者说，他的朋友保罗·艾伦看到了这本杂志，并把它拿给盖茨看，两人立即决定要为这台机器编写一个 BASIC 编译器。88^无论是盖茨还是罗伯茨决定为 Altair 使用 BASIC，最终都是 BASIC（图 7.6）。

William Gates III was born in 1955, at a time when work on FORTRAN was just underway. He was a student at Harvard when the famous cover of Popular Electronics appeared describing the Altair. According to one biographer, his friend Paul Allen saw the magazine and showed it to Gates, and the two immediately decided that they would write a BASIC compiler for the machine.⁸⁸ Whether it was Gates’s or Roberts’s decision to go with BASIC for the Altair, BASIC it was (figure 7.6).

在发给 Altair 客户的一份简报中，盖茨和艾伦表示，只需要 4K 字节内存的 BASIC 版本将于 1975 年 6 月上市，功能更强大的版本也将很快上市。对于那些同时购买 Altair 内存板的人来说，4K BASIC 的价格为 60 美元，8K 的价格为 75 美元，“扩展”BASIC（需要磁盘或其他大容量存储）的价格为 150 美元。那些希望该语言在另一个基于 8080 的系统上运行的人必须支付 500 美元^。89

In a newsletter sent out to Altair customers, Gates and Allen stated that a version of BASIC that required only 4K bytes of memory would be available in June 1975, and that more powerful versions would be available soon after. The cost, for those who also purchased Altair memory boards, was $60 for 4K BASIC, $75 for 8K, and $150 for “extended” BASIC (requiring disk or other mass storage). Those who wanted the language to run on another 8080-based system had to pay $500.⁸⁹

在蒙特·大卫多夫 (Monte Davidoff) 的帮助下，盖茨和艾伦精力充沛，不仅编写了占用极小内存的 BASIC，还编写了功能众多、性能卓越的 BASIC。该语言忠于其达特茅斯根源，易于学习。它打破了达特茅斯根源，提供了一种从 BASIC 命令转换为机器语言指令的方法。这主要是通过 USR 命令实现的，该命令借用自为 DEC 小型计算机编写的软件（其首字母缩写代表用户服务程序）。90^程序员甚至可以通过 PEEK 和 POKE 命令直接将字节放入特定内存位置或从特定内存位置提取数据，这会对分时达特茅斯系统造成严重破坏。与 USR 一样，这些命令也源自 DEC 程序员的先前工作，他们为 PDP-11 用 BASIC 编写的分时系统想出了这些命令。这些命令允许用户轻松地从 BASIC 转换为机器语言，这是让小型系统完成有用工作的关键特性。

In a burst of energy, Gates and Allen, with the help of Monte Davidoff, wrote not only a BASIC that fit into very little memory; they wrote a BASIC with a lot of features and impressive performance. The language was true to its Dartmouth roots in that it was easy to learn. It broke with those roots by providing a way to move from BASIC commands to instructions written in machine language. That was primarily through a USR command, which was borrowed from software written for DEC minicomputers (where the acronym stood for user service routine).⁹⁰ A programmer could even directly put bytes into or pull data out of specific memory locations, through the PEEK and POKE commands— which would have caused havoc on the time-shared Dartmouth system. Like USR, these commands were also derived from prior work done by DEC programmers, who came up with them for a time-sharing system they wrote in BASIC for the PDP-11. Those commands allowed users to pass from BASIC to machine language easily—a crucial feature for getting a small system to do useful work.

图 7.6

史密森尼收藏的包含 BASIC 1.1 版的纸带。根据比尔·盖茨在 1975 年 12 月出版的 Altair 用户组简报《计算机笔记》中的一封信：“如果有人在使用 BASIC 1.1 版，那么你手上就有一份 3 月份被盗的磁带副本。没有客户收到过 1.1 版，因为它是实验性的，而且漏洞百出！”（来源：史密森尼学会。）

Figure 7.6

Paper tape containing BASIC, version 1.1, from the Smithsonian Collections. According to a letter by Bill Gates in the December 1975 issue of the Altair Users Group newsletter, Computer Notes: “If anyone is using BASIC version 1.1, you have a copy of a tape that was stolen back in March. No customers were ever shipped 1.1, as it was experimental and is full of bugs!” (Source: Smithsonian Institution.)

这些扩展使 BASIC 保持在内存限制内，同时赋予它更复杂语言的性能。但它仍然是一种新手可以学习和使用的交互式对话语言。他们为 Altair 编写的 BASIC 巧妙地结合了达特茅斯和数字设备公司的功能，是盖茨和艾伦成功建立个人计算机软件行业的关键。

These extensions kept their BASIC within its memory constraints while giving it the performance of a more sophisticated language. Yet it remained an interactive, conversational language that novices could learn and use. The BASIC they wrote for the Altair, with its skillful combination of features taken from Dartmouth and from the Digital Equipment Corporation, was the key to Gates’s and Allen’s success in establishing a personal computer software industry.

与 Kemeny 和 Kurtz 不同，该语言的开发者并未接受过正式的计算机科学或数学培训。他们接触计算的方式略有不同。1968 年，比尔盖茨在西雅图的私立学校为学生提供了通用电气分时系统，当时即使是大学也很少有学生能使用该系统。后来，他接触到了更好的分时系统：计算机中心公司拥有的 PDP-10。后来，他使用一套 PDP-10 和 PDP-11 系统来控制邦纳维尔电力管理局的水力发电。他在邦纳维尔电力的导师之一是约翰诺顿，他是 TRW 的一名员工，曾参与阿波罗计划，因其工作质量而在程序员中享有传奇地位^。91

The developers of this language were not formally trained in computer science or mathematics as were Kemeny and Kurtz. They were introduced to computing in a somewhat different way. Bill Gates’s private school in Seattle had a General Electric time-sharing system available for its pupils in 1968, a time when few students even in universities had such access. Later on he had access to an even better time-shared system: a PDP-10 owned by the Computer Center Corporation. Later still, he worked with a system of PDP-10s and PDP-11s used to control hydroelectric power for the Bonneville Power Administration. One of his mentors at Bonneville Power was John Norton, a TRW employee who had worked on the Apollo Program and who was a legend among programmers for the quality of his work.⁹¹

盖茨在哈佛大学为 Altair 编写 BASIC 时，他没有接触过基于 8080 的系统，但他接触过哈佛大学计算中心的一台 PDP-10（以霍华德·艾肯命名）。他和同学蒙特·大卫多夫使用 PDP-10 编写了基于英特尔 8080 的书面规范的语言。1975 年初，保罗·艾伦飞往阿尔伯克基，向罗伯茨和耶茨演示了该语言。它成功了。不久之后，MITS 宣传了该语言可用于 Altair。其他人也在为 Altair 和当时充斥市场的其他小型计算机编写 BASIC 解释器，但没有一个能像盖茨和艾伦的解释器那样好，而且消息很快就传开了。

When he was writing BASIC for the Altair, Gates was at Harvard. He did not have access to an 8080-based system, but he did have access to a PDP-10 at Harvard’s computing center (named after Howard Aiken). He and fellow student Monte Davidoff used the PDP-10 to write the language, based on the written specifications of the Intel 8080. In early 1975 Paul Allen flew to Albuquerque and demonstrated it to Roberts and Yates. It worked. Soon after, MITS advertised its availability for the Altair. Others were also writing BASIC interpreters for the Altair and for the other small computers now flooding the market, but none was as good as Gates’s and Allen’s, and it was not long before word of that got around.

罗伯茨和他的公司似乎做出了一个又一个明智的决定：8080 处理器、总线架构，现在又有了 BASIC。然而，到 1975 年底，盖茨和艾伦却不这么认为。盖茨坚称自己从未成为 MITS 的员工（尽管艾伦直到 1976 年才成为员工），并且他和艾伦以“Micro Soft”（后来的“Micro-Soft”）的名义保留了 BASIC 的版权。92^在1976 年初发布的一封现已成为传奇的“致业余爱好者的公开信”中，盖茨抱怨人们通过复制纸带来非法复制他的 BASIC。盖茨声称“我们（开发该语言）所花费的计算机时间价值超过 40,000 美元。”他说，如果他和他的程序员得不到报酬，他们就没有动力为个人电脑开发更多软件，例如为 8080 处理器开发 APL 语言。他认为非法复制会使所有个人计算机面临风险：“没有什么比雇佣十名程序员并用好的软件充斥业余爱好者市场更让我高兴的了。” ⁹³

It seemed that Roberts and his company had made one brilliant decision after another: the 8080 processor, the bus architecture, and now BASIC. However, by late 1975 Gates and Allen were not seeing it that way. Gates insists that he never became a MITS employee (although Allen was until 1976), and that under the name “Micro Soft,” later “Micro-Soft,” he and Allen retained the rights to their BASIC.⁹² In a now-legendary “Open Letter to Hobbyists,” distributed in early 1976, Gates complained about people making illicit copies of his BASIC by duplicating the paper tape. Gates claimed “the value of the computer time we have used [to develop the language] exceeds $40,000.” He said that if he and his programmers were not paid, they would have little incentive to develop more software for personal computers, such as an APL language for the 8080 processor. He argued that illicit copying put all personal computing at risk: “Nothing would please me more than to hire ten programmers and deluge the hobby market with good software.”⁹³

盖茨在哈佛大学读本科时，就已经开始在 PDP-10 上工作。学生们不得将这台计算机用于商业用途，尽管当时这些区别并不像后来那么明显。该语言本身是达特茅斯学院的 Kemeney 和 Kurtz 发明的；对其成功至关重要的扩展来自数字设备公司的程序员，尤其是 Mark Bramhall，他领导了为 PDP-11 开发分时系统 (RSTS-11) 的努力。数字设备公司是上述集团中唯一的商业实体，它并不认为其软件是一种要出售的商品；它只是公司让人们购买硬件的行为^。94

Gates did his initial work on the PDP-10 while still an undergraduate at Harvard. Students were not to use that computer for commercial purposes, although these distinctions were not as clear then as they would be later. The language itself was the invention of Kemeney and Kurtz of Dartmouth; the extensions that were crucial to its success came from programmers at the Digital Equipment Corporation, especially Mark Bramhall, who led the effort to develop a time-sharing system (RSTS-11) for the PDP-11. Digital, the only commercial entity among the above group, did not think of its software as a commodity to sell; it was what the company did to get people to buy hardware.⁹⁴

比尔·盖茨认识到了罗伯茨和其他人没有认识到的事情：随着廉价个人电脑的出现，软件能够也应该成为计算的主要推动力。而且，只有收费才能实现这一点——尽管软件最初是免费的。到 1978 年，他的公司（现在称为“微软”）已经与 MITS 公司断绝关系，并从阿尔伯克基迁往西雅图郊区贝尔维尤。（MITS 公司本身已失去其身份，于 1977 年被 Pertec 公司收购。）正如斯图尔特·布兰德在 1972 年所预测的那样，计算机确实正在走向“人民”。但驱动力不是反主流文化的共享和免费信息乌托邦愿景，而是市场的力量。盖茨兑现了他的承诺，“雇佣十名程序员，淹没……市场”（图 7.7）。

Bill Gates had recognized what Roberts and all the others had not: that with the advent of cheap, personal computers, software could and should come to the fore as the principal driving agent in computing. And only by charging money for it—even though it had originally been free—could that happen. By 1978 his company, now called “Microsoft,” had severed its relationship with MITS and was moving from Albuquerque to the Seattle suburb of Bellevue. (MITS itself had lost its identity, having been bought by Pertec in 1977.) Computers were indeed coming to “the people,” as Stewart Brand had predicted in 1972. But the driving force was not the counterculture vision of a Utopia of shared and free information; it was the force of the marketplace. Gates made good on his promise to “hire ten programmers and deluge the ... market” (figure 7.7).

系统软件：拼图的最后一块

System Software: The Final Piece of the Puzzle

加里·基尔多尔 (Gary Kildall) 进入个人计算机软件领域是担任英特尔的顾问，他开发了系统开发语言。在从事这项工作期间，他意识到如果能够适当改造，软盘将成为小型系统的良好大容量存储设备。为此，他编写了一个小程序来管理软盘驱动器的信息流。与选择 BASIC 一样，事后看来，软盘成为个人计算机的大容量存储介质是显而易见且不可避免的。这忽略了它从未打算用于此用途的事实。与 BASIC 的改造一样，软盘必须重新塑造为新的角色。与 BASIC 一样，这项工作需要许多人的努力，但主要的努力来自一个人，加里·基尔多尔 (Gary Kildall)。

Gary Kildall’s entree into personal computing software was as a consultant for Intel, where he developed languages for system development. While doing that he recognized that the floppy disk would make a good mass storage device for small systems, if it could be properly adapted. To do that he wrote a small program that managed the flow of information to and from a floppy disk drive. As with the selection of BASIC, it appears in hindsight to be obvious and inevitable that the floppy disk would be the personal computer’s mass storage medium. That ignores the fact that it was never intended for that use. As with the adaptation of BASIC, the floppy had to be recast into a new role. As with BASIC, doing that took the work of a number of individuals, but the primary effort came from one man, Gary Kildall.

磁盘比磁带或纸带有几个优势。首先，它速度更快。其次，用户可以在磁盘上读取和写入数据。它的主要优势是磁盘具有“随机”访问：用户不必浏览整个磁带卷轴即可获取特定数据。然而，要做到这一点需要复杂的编程——IBM 在其大型机系统中称之为磁盘操作系统或^DOS。95

A disk had several advantages over magnetic or paper tape. For one, it was faster. For another, users could both read and write data on it. Its primary advantage was that a disk had “random” access: Users did not have to run through the entire spool of tape to get at a specific piece of data. To accomplish this, however, required tricky programming—some thing IBM had called, for one of its mainframe systems, a Disk Operating System, or DOS.⁹⁵

图 7.7

微软团队，约 1978 年。这张照片显示微软正从 Altair 的制造地阿尔伯克基迁往比尔盖茨（左下）和保罗艾伦（右下）的家乡西雅图地区。当时微软还是一家小公司，主要专注于为个人电脑提供编程语言。（来源：微软。）

Figure 7.7

Microsoft Team, ca. 1978. This photograph shows Microsoft as it was moving from Albuquerque, where the Altair was built, to the Seattle area, where Bill Gates (lower left) and Paul Allen (lower right) were from. It was still a small company that focused mainly on supplying programming languages for personal computers. (Source: Microsoft.)

个人计算机 DOS 与大型机操作系统关系不大。它无需安排和协调许多用户的工作：Altair 只有一个用户。它无需“假脱机”或以其他方式将数据传送到满屋子的链式打印机、打卡机和磁带驱动器：个人计算机只需要担心几个端口。它需要的是快速准确地存储和检索软盘中的文件。事实上，典型的文件将存储为一组碎片，插入磁盘上可用的任何可用空间。操作系统的工作是找到这些可用空间，将数据放入其中，稍后检索数据，然后重新组装碎片。所有这些都给用户一种错觉，认为磁盘就像一个传统的文件柜，里面装满了包含纸质文件的文件夹。

A personal computer DOS had little to do with mainframe operating systems. There was no need to schedule and coordinate the jobs of many users: an Altair had one user. There was no need to “spool” or otherwise direct data to a roomful of chain printers, card punches, and tape drives: a personal computer had only a couple of ports to worry about. What was needed was rapid and accurate storage and retrieval of files from a floppy disk. A typical file would, in fact, be stored as a set of fragments, inserted at whatever free spaces were available on the disk. It was the job of the operating system to find those free spaces, put data there, retrieve it later on, and reassemble the fragments. All that gave the user an illusion that the disk was just like a traditional file cabinet filled with folders containing paper files.

再一次，数字设备公司成为先驱，部分原因是它的文化；因为它的许多员工都曾在麻省理工学院使用过 TX-0，这是第一批具有对话式互动感觉的计算机之一。DEC 在其早期系统中引入了 DECtape，虽然它是磁带，但允许程序员快速访问卷轴中间和末端的数据。96 ^PDP10具有强大的 DECtape 和磁盘存储能力；它的操作系统在创造个人计算的幻觉方面发挥了关键作用，这给斯图尔特·布兰德等观察家留下了深刻印象。

Once again, Digital Equipment Corporation was the pioneer, in part because of its culture; because of the experience many of its employees had had with the TX-0 at MIT, one of the first computers to have a conversational, interactive feel to it. For its early systems DEC introduced DECtape, which although a tape, allowed programmers rapid access to data written in the middle, as well as at the ends, of the reel.⁹⁶ The PDP10s had powerful DECtape as well as disk storage abilities; its operating systems were crucial in creating the illusion of personal computing that had so impressed observers like Stewart Brand.

20 世纪 60 年代末，DEC 为 PDP-8 开发了 OS/8，它具有 PDP-10 的感觉，但运行在内存非常有限的机器上。OS-8 让 DEC 的每个人都大开眼界；它表明小型计算机可以拥有与大型机一样复杂的功能，而没有大型机系统软件的臃肿。PDP11 的高级版本有一个名为 RT-11（于 1974 年推出）的操作系统，它类似于 OS/8，并进一步完善了管理磁盘数据的概念。97^这些是个人计算机操作系统的根源。DEC 在创建此软件方面的作用与其发明小型计算机一样，对个人计算的创造做出了重大贡献。

In the late 1960s DEC produced OS/8 for the PDP-8, which had the feel of the PDP-10 but ran on a machine with very limited memory. OS-8 opened everyone’s eyes at DEC; it showed that small computers could have capabilities as sophisticated as mainframes, without the bloat that characterized mainframe system software. Advanced versions of the PDP11 had an operating system called RT-11 (offered in 1974), which was similar to OS/8, and which further refined the concept of managing data on disks.⁹⁷ These were the roots of personal computer operating systems. DEC’s role in creating this software ranks with its invention of the minicomputer as major contributions to the creation of personal computing.

加里·基尔多尔 (Gary Kildall) 为 Intel 8080 开发了 PL/M。他使用的是 IBM System/360，PL/M 与 IBM 的 PL/I 类似。在从事该项目期间，基尔多尔为大型机的磁盘驱动器编写了一个小型控制程序。“事实证明，幸运的是，该操作系统（称为 CP/M，即微机控制程序）也很有用。” ⁹⁸基尔多尔说，PL/M 是“CP/M 的基础”，尽管这些命令显然源自 Digital 的软件，而不是 IBM 的软件。99^例如，用字母指定正在使用的驱动器；给文件名添加句点和三个字符的扩展名；使用 DIR（目录）命令、PIP 和 DDT 是 DEC 功能，这些功能均未经更改地保留下来。¹⁰⁰ 1976 年 4 月，吉姆·沃伦 (Jim Warren) 在《Dr. Dobb's Journal of Computer Calisthenics and Orthodontia [原文如此]》上发表文章，向业余爱好者宣布 CP/M“类似于 DECSYSTEM 10”。沃伦对 CP/M 感到兴奋，称它“设计精良，基于一个已经存在十年的易于使用的操作系统”。[原文如此] ¹⁰¹建议价格远低于 100 美元，完整的软盘系统（包括驱动器和控制器）售价约为 800 美元——不便宜，但明显优于盒式磁带、纸带或任何其他形式的磁带。CP/M 是拼图的最后一块，一旦面世，个人电脑就成为现实。

Gary Kildall developed PL/M for the Intel 8080. He used an IBM System/360, and PL/M was similar to IBM’s PL/I. While working on that project Kildall wrote a small control program for the mainframe’s disk drive. “It turned out that the operating system, which was called CP/M for Control Program for Micros, was useful, too, fortunately.”⁹⁸ Kildall said that PL/M was “the base for CP/M,” even though the commands were clearly derived from Digital’s, not IBM’s software.⁹⁹ For example, specifying the drive in use by a letter; giving file names a period and three-character extension; and using the DIR (Directory) command, PIP, and DDT were DEC features carried over without change.¹⁰⁰ CP/M was announced to hobbyists as “similar to DECSYSTEM 10” in an article by Jim Warren in Dr. Dobb’s Journal of Computer Calisthenics and Orthodontia [sic] in April 1976. Warren was excited by CP/M, stating that it was “well designed, based on an easy-to-use operating system that has been around for a DECade.[sic]”¹⁰¹ Suggested prices were well under $100, with a complete floppy system that included a drive and a controller for around $800—not cheap, but clearly superior to the alternatives of cassette, paper tape, or any other form of tape. CP/M was the final piece of the puzzle that, when made available, made personal computers a practical reality.

加里·基尔多尔和他的妻子多萝西·麦克尤恩在加利福尼亚州蒙特雷的海军研究生院担任教员期间，逐渐进入了商业软件行业（图 7.8 ）。随着人们对 CP/M 的兴趣日渐增加，他开始为其他客户编写 CP/M 的变体。《多布博士的日记》上的宣传带来了足够的销量，使他相信 CP/M 的潜在市场。1976 年，他辞去了工作，与多萝西创办了一家名为 Digital Research（最初名为 Intergalactic Digital Research）的公司，其主要产品是 CP/ ^M。102

Gary Kildall and his wife, Dorothy McEwen, eased themselves into the commercial software business while he also worked as an instructor at the Naval Postgraduate School in Monterey, California (figure 7.8). As interest in CP/M picked up, he found himself writing variations of it for other customers. The publicity in Dr. Dobb’s Journal led to enough sales to convince him of the potential market for CP/M. In 1976 he quit his job and with Dorothy founded a company, Digital Research (initially Intergalactic Digital Research), whose main product was CP/M.¹⁰²

第二年，也就是 1977 年，他设计了一个具有重要区别的版本。IMSAI 是一家生产了 Altair“克隆版”（图 7.9）的公司，该公司希望获得在其产品上使用 CP/M 的许可。在与 IMSAI 员工 Glen Ewing 合作的过程中，Kildall 重写了 CP/M，这样只有一小部分需要根据 IMSAI 的具体情况进行定制。其余部分将是通用代码，不必在每次出现新计算机或磁盘驱动器时重写。他将专门的代码称为 BIOS（基本输入/输出系统）。103^这一变化使系统软件标准化，就像 100 针 Altair 总线标准化硬件一样。IMSAI 的计算机系统成为了标准，它具有坚固的电源、大量内部插槽的扩展空间、外部软盘驱动器和 CP/M。

The next year, 1977, he designed a version with an important difference. IMSAI, the company that had built a “clone” of the Altair (figure 7.9), wanted a license to use CP/M for its products. Working with IMSAI employee Glen Ewing, Kildall rewrote CP/M so that only a small portion of it needed to be customized for the specifics of the IMSAI. The rest would be common code that would not have to be rewritten each time a new computer or disk drive came along. He called the specialized code the BIOS—Basic Input/Output System.¹⁰³ This change standardized the system software in the same way that the 100-pin Altair bus had standardized hardware. IMSAI’s computer system became a standard, with its rugged power supply, room for expansion with plenty of internal slots, external floppy drive, and CP/M.

图 7.8

Gary Kildall。背景中可见 DEC VT-100 终端。（来源： Kristen Kildall。）

Figure 7.8

Gary Kildall. A DEC VT-100 terminal is visible in the background. (Source: Kristen Kildall.)

图 7.9

IMSAI 8080，Altair 最成功的复制品之一，配有 Micropolis 提供的视频监视器和磁盘存储系统。（来源：史密森学会。）

Figure 7.9

IMSAI 8080, one of the most successful copies of the Altair, with a video monitor and a disk storage system supplied by Micropolis. (Source: Smithsonian Institution.)

开拓阶段结束，1977 年

End of the Pioneering Phase, 1977

到1977 年，所有部件都已就位。Altair 的设计缺陷已得到纠正，如果不是 MITS 公司，就是其他公司。Microsoft BASIC 允许程序员为这些机器编写有趣且首次严肃的软件。这种软件收费的道德规范激励了这些程序员，尽管软件盗版也已成常态。计算机也提供只读存储器 (ROM) 上的 BASIC，制造商只需向 Microsoft 支付一笔简单的版税。（由于启动代码也在 ROM 中，因此不再需要带有指示灯和开关阵列的前面板。）由 CP/M 控制的 8 英寸软盘驱动器提供了一种独立于特定型号的开发和交换软件的方法。机器配备了标准化的串行和并行端口，以及打印机、键盘和视频监视器的连接。最后，到 1977 年，出版、软件公司和支持团体等强大而健康的行业已经出现，让新手也能使用。个人电脑已经到来。

By 1977 the pieces were all in place. The Altair’s design shortcomings were corrected, if not by MITS then by other companies. Microsoft BASIC allowed programmers to write interesting and, for the first time, serious software for these machines. The ethic of charging money for this software gave an incentive to such programmers, although software piracy also became established. Computers were also being offered with BASIC supplied on a read-only-memory (ROM), the manufacturer paying Microsoft a simple royalty fee. (With the start-up codes also in ROM, there was no longer a need for the front panel, with its array of lights and switches.) Eight-inch floppy disk drives, controlled by CP/M, provided a way to develop and exchange software that was independent of particular models. Machines came with standardized serial and parallel ports, and connections for printers, keyboards, and video monitors. Finally, by 1977 there was a strong and healthy industry of publications, software companies, and support groups to bring the novice on board. The personal computer had arrived.

增强人类智力，1975-1985

Augmenting Human Intellect, 1975–1985

20 世纪 70 年代中期，在小型系统领域草根性的活力和创造力中，还发生了什么？当老牌计算机公司看到个人计算机出现时，它们也进入了创造力和技术进步的时期。起初几乎没有重叠。但到了 1985 年，重叠越来越多：基于廉价微处理器的个人计算模式强行进入了该行业。本章将介绍这种情况是如何发生的。

In the mid-1970s, amid the grassroots energy and creativity of the small systems world, what else was happening? When the established computer companies saw personal computers appear, they, too, entered a period of creativity and technical advance. At first there was little overlap. By 1985, though, there was overlap and more: the paradigm of personal computing based on inexpensive microprocessors forced itself onto the industry. This chapter looks at how that happened.

数字设备公司

Digital Equipment Corporation

数字设备公司 (DEC) 凭借其小型计算机和软件为交互式个人计算奠定了基础。当英特尔宣布推出 8080（一款在一块芯片上具备小型计算机基本功能的设备）时，他们在做什么？“我们正处于打造 VAX 的阵痛之中。” ¹ VAX 是 PDP-11 的扩展，可达到大型机的性能。对于 DEC 来说，这是一项重大任务，占用了公司的资源。正如 IBM 对其 System/360 所做的那样，DEC 将“公司赌注”押在了 VAX 上 — 迈向更高性能和更大型系统。

Digital Equipment Corporation built the foundation for interactive personal computing with its minicomputers and its software. What were they doing when Intel announced its 8080, a device with the essentials of a minicomputer on one chip? “We were just in the throes of building the VAX.”¹ The VAX was an extension of the PDP-11 that reached toward mainframe performance. It was a major undertaking for DEC and strained the company’s resources. As IBM had done with its System/360, Digital “bet the company” on the VAX—a move toward higher performance and larger systems.

DEC 内部的许多人都认为，与其说该公司是一家小型计算机制造商，不如说它是一家销售架构的公司。2^从TX-0 开始，DEC 的创始人就一直以他们能够通过创新设计制造高性能计算机（无论大小）的能力而自豪。这也许可以解释为什么 DEC 未能应对英特尔等公司对其业务构成的威胁。围绕英特尔 8080 制造计算机意味着将架构决定权交给半导体公司——他们怎么能允许自己这样做呢？另一种选择是将 PDP-11 指令集授权给芯片制造商，由他们生产基于该指令集的微处理器，这种选择同样遭到拒绝。该公司认为，这样做会将“公司珍宝”拱手相让。Digital 确实在 1974 年生产了 LSI-11，即单板 PDP-11，但这并没有带来像英特尔 8080 那样廉价的系统。Digital 开发了名为 T-11 的单芯片 PDP-11，但从未上市。尽管 PDP-11 架构的元素曾出现在微处理器设计中（尤其是摩托罗拉 6800），但微处理器现象却与 PDP-11 擦肩而过^。3

Many within DEC felt that the company was not so much a minicomputer builder as it was a company that sold architecture.² Beginning with the TX-0, DEC’s founders had taken pride in their ability to build high-performance computers—large or small—through innovative design. That may explain why DEC failed to counter the threat that companies like Intel posed to its business. To build a computer around the Intel 8080 meant surrendering decisions about architecture to a semiconductor house—how could they allow themselves to do that? The other alternative, licensing the PDP-11 instruction set to chip makers, who would produce microprocessors based on it, was likewise rejected. The company thought that would be giving the “corporate jewels” away. Digital did produce the LSI-11, a single-board PDP-11, in 1974, but that did not lead to inexpensive systems as did the Intel 8080. A single-chip PDP-11, called T-11, was developed but never marketed. The microprocessor phenomenon passed the PDP-11 by, even though elements of its architecture turned up in microprocessor designs (especially the Motorola 6800).³

PDP-11 扩展计划始于 1974 年或 1975 年。1977 年 10 月，DEC 发布了 VAX，型号 11/780（图 8.1）。全名是 VAX-11，代表 PDP-11 的虚拟地址扩展。这意味着 VAX 只是一个地址空间为 32 位而不是 16 位的 PDP-11。事实上，VAX 确实是一台新机器。但是，它可以通过设置调用 PDP-11 指令集的“模式位”来执行现有的 PDP-11 软件。（最终放弃了兼容模式。）

Planning for an extension to the PDP-11 began in 1974 or 1975. DEC announced the VAX, Model 11/780, in October 1977 (figure 8.1). The full name was VAX-11, which stood for Virtual Address eXtension [of the] PDP-11. The implication was that the VAX was simply a PDP-11 with a 32-bit instead of a 16-bit address space. In fact, the VAX was really a new machine. It could, however, execute existing PDP-11 software by setting a “mode bit” that called forth the PDP-11 instruction set. (Eventually the compatibility mode was dropped.)

DEC 继续以更低的价格和更小的封装尺寸销售小型计算机，例如 1975 年推出的售价不到 3,000 美元的 PDP-8/A。4^但该公司更倾向于开发和销售更高性能的计算机。它给出的一个原因是，对于给定的应用程序，计算机的成本只是总成本的一部分；还有“与 [现有] 应用程序相关的高固定间接成本”。5^显然，DEC 并不认为它能够像 MITS 对 Altair 所做的那样实现真正大幅的降价。这个理由，再加上 DEC 不愿将其计算机架构技术转让给半导体公司，使得该公司在 1974 年至 1977 年这个最容易进入个人计算机市场的关键时期未能进入该市场。

DEC continued to market small computers at successively lower prices and in smaller packages, for example, the PDP-8/A, introduced in 1975 for under $3,000.⁴ But the company preferred to develop and market higher performance. One reason it gave was that for a given application, the cost of the computer was only part of the total cost; there was also “the high fixed overhead costs associated with the [existing] application.”⁵ Apparently DEC did not feel it could achieve truly drastic price reductions, as MITS had done with the Altair. That argument, coupled with DEC’s reluctance to turn over its skill in computer architecture to semiconductor companies, kept the company out of the personal computer market during the crucial years, 1974 to 1977, when it could most easily have entered it.

图 8.1

VAX 11/780，约 1978 年。（来源：数字设备公司。）

Figure 8.1

VAX 11/780, ca. 1978. (Source: Digital Equipment Corporation.)

正如 DEC 并非第一个推出 16 位小型机的公司一样，它也不是第一个将地址空间扩展到 16 位以上的公司。1973 年，同样位于马萨诸塞州 128 号公路旁的 Prime 公司推出了一台 32 位小型计算机。Prime 发展迅速，直到 20 世纪 80 年代末与 Computervision 合并。另一家公司 Interdata 于 1974 年描述了一款“超级小型机”。他们的设计也获得了商业上的成功，同年该公司被康涅狄格州光学公司 Perkin-Elmer 收购。6^佛罗里达州劳德代尔堡的系统工程实验室 (Systems Engineering Laboratory) 也推出了一款 32 位小型机，受到 NASA 和航空航天客户的欢迎。SEL 于 1980 年被卖给 Gould，并成为这家老牌公司进军计算机业务的基础。7^这些发展的动力是越来越多的相对便宜的半导体存储器可用来替代磁芯。这些内存芯片使设计具有大容量主内存的机器变得更加实用，而这又需要更多的地址空间。

Just as DEC was not the first to market a 16-bit mini, it was not the first to extend address space beyond 16 bits. In 1973, Prime, also located off Route 128 in Massachusetts, shipped a 32-bit minicomputer. Prime grew rapidly until merging with Computervision in the late 1980s. Another company, Interdata, described a “mega-mini” in 1974. Their design was also commercially successful, and that year the company was bought by Perkin-Elmer, the Connecticut optics company.⁶ Systems Engineering Laboratory of Fort Lauderdale, Florida, also introduced a 32-bit mini, which was popular with NASA and aerospace customers. S.E.L. was sold to Gould in 1980 and became the basis for that venerable company’s entree into the computer business.⁷ The impetus for these developments was the growing availability of relatively cheap semiconductor memory to replace magnetic core. These memory chips made it more practical to design machines with large main memories, which in turn demanded more address space.

如果 VAX 只是名义上 PDP-11 的扩展，那么它实际上是一台“虚拟”内存计算机。该术语的非正式定义是，它是一种使计算机小而快的主内存看起来比实际更大的方法，通过在磁盘上较慢但较大的内存之间交换数据。更精确的定义涉及执行此操作的方式：首先，此过程不能严重降低整体性能，其次，用户不必知道正在进行这种交换（因此有这样的术语：内存“虚拟”很大但实际上很小）^。8

If the VAX was only nominally an extension of the PDP-11, it was genuinely a “virtual” memory computer. An informal definition of this term is that it is a way to make a computer’s small but fast main memory seem to be bigger than it is, by swapping data to and from a slower but larger memory on a disk. A more precise definition concerns the way this is done: first of all, overall performance must not be seriously degraded by this process, and second, the user should not have to know that this swapping is going on (hence the term: the memory is “virtually” large but in reality small).⁸

20 世纪 40 年代末，伯克斯、戈德斯坦和冯·诺依曼在高等研究院的报告中讨论了对内存层次的需求，每个内存都比下面的内存慢，但内存容量更大。Atlas 由英国曼彻斯特大学设计，费兰蒂于 1962 年制造，可能是第一个使用这种设计的计算机，它给用户一种单级快速大容量内存的错觉。9^它是当时世界上最快的计算机之一，也是对后续几代计算机影响最大的计算机之一。Atlas 的用户看到了一台虚拟内存为 100 万个 48 位字的机器。计算机根据一组寄存器的内容自动在核心和磁鼓之间交换数据（一种称为关联内存寻址的技术）。10^尽管Atlas 的商业版本影响深远，但对费兰蒂来说，它只是取得了适度的成功。在美国，Burroughs 在 20 世纪 60 年代中期提供了虚拟内存，并在架构上取得了一些重要的进步。 IBM 于 1972 年推出的 System/370 型号中就提供了该内存。（虚拟内存这个术语可能就是从 370 的营销开始被广泛使用的。）¹¹ SDS 940 分时系统也遵循了 Atlas 的设计。

The need for a hierarchy of memories, each slower but larger than the one below it, was discussed in the Institute for Advanced Study reports by Burks, Goldstine, and von Neumann in the late 1940s. The Atlas, designed at Manchester University in England and built by Ferranti in 1962, was probably the first to use a design that gave the user the illusion of a single-level fast memory of large capacity.⁹ It was one of the fastest computers in the world at the time and also one of the most influential on successive generations. A user of the Atlas saw a machine with a virtual memory of one million 48-bit words. The computer automatically swapped data between the core and the drum, based on the contents of a set of registers (a technique called associative memory-addressing).¹⁰ Though influential, commercial versions of the Atlas were only a modest success for Ferranti. In the United States, Burroughs offered virtual memory, with some important architectural advances, in the mid-1960s. IBM offered it with System/370 models announced in 1972. (It is probably from marketing the 370 that the term virtual memory came into wide use.)¹¹ The SDS 940 time-sharing system also followed the Atlas design.

C. Gordon Bell 领导了 VAX 的初始设计工作，Bill Strecker 是其首席架构师。突破 PDP11 65 KB 地址空间的限制是他们的主要目标。VAX 提供了₂ 32 或 4.3 GB（相当于十亿个 32 位字）的虚拟地址空间。其寻址方案将内存划分为块（称为页面），并使用关联比较来确定所需页面是否在核心中。VAX 处理器使用 16 个 32 位通用寄存器，就像 IBM 360 一样。它还拥有一套丰富的指令，超过 250 条，具有九种不同的寻址模式，允许一条指令执行复杂的操作^。12

C. Gordon Bell led the initial design effort for the VAX, and Bill Strecker was its chief architect. Breaking through the limits of the PDP11’s 65 Kbyte address space was their primary goal. The VAX provided ₂32 or 4.3 gigabytes (equivalent to one billion 32-bit words) of virtual address space. Its addressing scheme divided memory into blocks, called pages, and used an associative comparison to determine whether the desired page was in core or not. The VAX processor used sixteen 32-bit general registers, like the IBM 360. It also had a rich set of over 250 instructions with nine different addressing modes, which allowed a single instruction to carry out complex operations.¹²

VAX 取得了商业上的成功，在接下来的十年里销售了大约 100,000 台，尽管它出现得晚，但还是超越了其他 32 位小型机。11/780 的性能大致为每秒 100 万条指令 (MIPS)，成为竞争对手在 20 世纪 90 年代比较其机器的基准。随后出现了整个“Vaxen”系列：1980 年推出的功能较弱的 11/750、1984 年推出的性能较高的 8600 以及 1985 年推出的紧凑型 MicroVax II 等。13这些机器让 DEC 保持盈利并在 128 号公路上占据主导地位^。即使是 Data General（其 Nova 曾是 PDP-11 的强大竞争对手）也难以与 VAX 竞争，尽管它确实在 1980 年推出了 32 位 Eclipse，正如 Tracy Kidder 的畅销书《新机器的灵魂》中所记载的^那样。14

The VAX was a commercial success, selling around 100,000 over the next decade and leaping over the other 32-bit minis even though it appeared later. The 11/780’s performance, roughly calculated at one million instructions per second (MIPS), became a benchmark against which competitors would compare their machines into the 1990s. A whole family of “Vaxen” followed: the less-powerful 11/750 in 1980, the higher-performance 8600 in 1984, and the compact MicroVax II in 1985, among others.¹³ These machines kept DEC profitable and dominant along Route 128. Even Data General, whose Nova had been such a strong competitor for the PDP-11, had trouble competing with the VAX, although it did introduce a 32-bit Eclipse in 1980, as chronicled in Tracy Kidder’s bestseller The Soul of a New Machine.¹⁴

VAX 是一款通用计算机，配备标准语言和软件。它销往广泛的市场，但其最大的影响是在工程和科学领域。起价为 120,000 美元，对于单个工程师来说太贵了，但对于航空航天、汽车或化学公司的一个部门来说却足够便宜。对于他们来说，标准做法要么是排队使用公司的大型机，要么是签约使用商业分时服务。VAX 为他们提供了现成的计算能力。它有一个可靠的、面向工程的操作系统 (VMS) 和用于数据收集的复杂 I/O 设施。

The VAX was a general-purpose computer that came with the standard languages and software. It sold to a wide market, but its biggest impact was on engineering and science. Prices started at $120,000, which was too expensive for a single engineer, but just cheap enough to serve a division at an aerospace, automotive, or chemical firm. For them the standard practice had been either to get in line to use the company’s mainframe, or to sign up for time on a commercial time-sharing service. The VAX gave them computing power at hand. It had a solid, engineering-oriented operating system (VMS), and sophisticated I/O facilities for data collection.

最后，VAX 配备了功能强大且易于使用的终端 VT-100。它具有令人印象深刻的功能数量，但人们觉得没有一个是多余的。它设法保留了旧电传打字机的舒适感。许多用户喜欢的一项功能是它能够一次滚动一个像素，而不是一次一行。没有实际理由拥有此功能，它未能与其他终端显示器流行起来，但它具有很大的吸引力。VT-100 的代码使用 ASCII，确实成为未来二十年终端的标准。

Finally, the VAX came with a powerful and easy-to-use terminal, the VT-100. It had an impressive number of features, yet one felt that none was superfluous. It somehow managed to retain the comfortable feel of the old Teletype. One feature that many users loved was its ability to scroll a pixel at a time, rather than a line at a time. There was no practical reason to have this feature, and it failed to catch on with other terminal displays, but it had a great appeal. The VT-100’s codes, using ASCII, did become the standard for terminals for the next twenty years.

关于 UNIX

A Word about UNIX

下一章将更全面地讨论 UNIX 对商业计算的影响，这里我们仅简要介绍它相对于 VAX 的地位。除了 VMS，VAX 的 PDP-11 祖先意味着用户也可以在其上运行 UNIX。UNIX 是在 DEC 小型计算机上开发的，在 UNIX 诞生的最初几年里，它只在 DEC 计算机上运行，主要是 PDP-11。加州大学伯克利分校的 UNIX 版本对计算和互联网产生了巨大影响，它是在 VAX 上开发的。尽管如此，DEC 对 VAX 的 UNIX 持矛盾态度。据称 Ken Olsen 曾说过“UNIX 是骗人的鬼话！”（他发表该言论的上下文存在争议。15 ^）无论如何，VAX 可以并且确实运行了伯克利 UNIX，至少在形成时期，VAX 计算机是互联网上最常见的节点^。16

The impact of UNIX on commercial computing will be discussed more fully in the next chapter, and here we will just briefly describe its place with regard to the VAX. In addition to VMS, the VAX’s PDP-11 ancestry meant that users could also run UNIX on it. UNIX was developed on DEC minicomputers, and for the first few years of UNIX’s existence it ran only on DEC computers, mainly PDP-11s. The University of California at Berkeley’s version of UNIX, which had an enormous influence on computing and on the Internet, was developed on a VAX. Still, DEC was ambivalent about UNIX for the VAX. Ken Olsen allegedly stated at one point that “UNIX is snake oil!” (The context in which he made that statement has been disputed.¹⁵) At any rate, the VAX could and did run Berkeley UNIX, and for at least the formative years, VAX computers were the most common nodes on the Internet.¹⁶

IBM 与经典大型机文化

IBM and the Classic Mainframe Culture

20 世纪 70 年代中期，个人电脑正在发明，Digital 公司正在制造 VAX，而 IBM 在做什么呢？与 Digital 公司一样，IBM 也忙于扩展其现有产品线，高端的 3033 于 1977 年初发布，低价的 4300 系列于 1979 年发布。后一个系列的性价比比当时市场上的中档 370 系统大幅提升，这一改进主要源于使用大规模集成电路。这些 LSI 芯片由 IBM 内部开发和设计，与英特尔或 Fairchild 等公司销售的芯片并不相似。17^随着System/370 安装的数量和复杂性不断增加，互连问题也随之出现。IBM 的 Bob Evans 指出，在 20 世纪 70 年代初期，大量不兼容和临时的网络方案与 IBM 十年前试图减少的计算机架构混乱情况类似。¹⁸其结果是系统网络架构 (SNA)，于 1974 年首次发布。SNA 是一套详细的标准。它为 20 世纪 90 年代大型计算机系统的联网奠定了基础。

In the mid-1970s, while the personal computer was being invented and while Digital was building the VAX, what was IBM doing? Like Digital, IBM was busy extending its existing line, with the high-end 3033 announced in early 1977, and the low-priced 4300-series announced in 1979. This latter series offered a dramatic increase in performance per dollar over the mid-range 370 systems then being marketed, an improvement that came mainly from using large-scale integrated circuits. These LSI chips were developed and designed by IBM in-house and did not resemble the ones being marketed by companies like Intel or Fairchild.¹⁷ As System/370 installations grew in number and complexity, the issue of interconnecting them also arose. Bob Evans of IBM remarked that, in the early 1970s, the plethora of incompatible and ad hoc networking schemes resembled the chaos of computer architectures that IBM had sought to reduce a decade before.¹⁸ The result was Systems Network Architecture (SNA), first shipped in 1974. SNA was a layer cake of standards, spelled out in detail. It formed the basis for networking large computer systems into the 1990s.

1975 年，IBM 推出了一款可能与其大型机定位不一致的产品：一台“个人”计算机，型号 5100。这台机器可以放在桌子上，在一个包装内包含处理器、键盘、磁带驱动器和小型视频终端。一台拥有 16 KB 内存的机器起价为 9,000 美元。19^它支持 BASIC 和 APL（APL 的开发者 Kenneth Iverson 于 1960 年加入 IBM），用户可以通过拨动前面板上的开关来选择。但几乎没有可用的应用软件；围绕 Altair 成长起来的第三方支持社区未能为 5100 实现。销量不高但稳定。（稍后将讨论“另一款”IBM 个人计算机。）

In 1975 IBM introduced a product that might have seemed at odds with its mainframe orientation: a “personal” computer, Model 5100. This machine could fit on a desk and contained a processor, keyboard, cassette tape drive, and small video terminal in a single package. Prices began at $9,000 for a machine with 16 Kbytes of memory.¹⁹ It supported both BASIC and APL (the developer of APL, Kenneth Iverson, had joined IBM in 1960), which the user could select by flipping a switch on the front panel. But little or no applications software was available; the third-party support community that grew up around the Altair failed to materialize for the 5100. Sales were modest but steady. (The “other” IBM personal computer will be discussed shortly.)

对于 IBM 当时所做所为这个问题的另一个回答是，它当时在法庭上。对于 IBM 来说，20 世纪 70 年代是诉讼的十年：美国诉 IBM，于 1969 年 1 月 17 日提起，并于 1981 年驳回。指控称 IBM 因其在美国通用电子数字计算机市场的主导地位而违反了反垄断法。司法部根据“市场”的定义提出这一指控，该定义涵盖了大型计算机所提供的面向业务的电子数据处理活动，其中 IBM 占据了大约 70% 的市场，“BUNCH”占据了几乎所有剩余的市场。IBM 反驳说，它的竞争对手不仅是 Burroughs、Univac、NCR、CDC 和 Honeywell，而是成千上万家制造和销售计算机、外围设备、软件、服务等的大大小小的公司。经过漫长的调查过程，在此期间，大多数公司的代表都提供了证词，最终，该案于 1975 年 5 月开庭审理 - 当时，比尔·盖茨和保罗·艾伦正在讨论为 Altair 开发 BASIC。

Another answer to the question of what IBM was doing is that it was in court. For IBM the 1970s was the decade of the lawsuit: U.S. vs. IBM, filed January 17, 1969, and dismissed in 1981. The charge was that IBM was in violation of antitrust laws by virtue of its dominance of the U.S. market for general-purpose electronic digital computers. The Justice Department based this charge on a definition of “market” that covered the business-oriented electronic data-processing activities served by mainframe computers, of which IBM held about 70 percent of the market and the “BUNCH” nearly all the rest. IBM countered by arguing that its competition was not just Burroughs, Univac, NCR, CDC, and Honeywell, but rather thousands of companies, large and small, that made and sold computers, peripherals, software, services, and the like. After a long discovery process, during which depositions were taken from representatives of most of these companies, the case finally went to trial in May, 1975—that is, around the time that Bill Gates and Paul Allen were talking about developing BASIC for the Altair.

调查过程和证词非常详尽。证词和证词的记录长达数千页。20^但个人系统风波没有出现在审判中。比尔·盖茨和埃德·罗伯茨都没有被传唤作证或提供证词。法院将注意力集中在前“侏儒”身上，尤其是已经退出该行业的 RCA 和 GE。偶尔，也会审查在一两个地方与 IBM 大型机竞争的公司。其中包括 SDS（当时是施乐的子公司），其计算机在科学应用方面与 System/360 Model 44 竞争，以及数字设备公司，不是在小型机方面竞争，而是在 PDP-10 方面竞争。法院甚至仔细审查了老牌缝纫机公司 Singer，该公司于 1963 年收购了 Friden，并在销售点零售终端领域建立了业务。（英国公司 ICL 于 1976 年收购了 Singer 的计算业务。）

The discovery process and the testimony were thorough and detailed. Transcripts of the depositions and testimony run into thousands of pages. ²⁰ But none of the gathering storm of personal systems made it into the trial. Neither Bill Gates nor Ed Roberts was called to testify or give depositions. The court focused its attention on the former “Dwarfs,” especially RCA and GE, who had left the business. Occasionally firms that competed with IBM’s mainframes at one or two places were examined. These included SDS (a subsidiary of Xerox by then), whose computers competed with the System/360 Model 44 for scientific applications, and Digital Equipment Corporation, not for its minicomputers but for its PDP-10. The court even looked closely at Singer, the venerable sewing machine company, which had purchased Friden in 1963 and built up a business in point-of-sale retail terminals. (The British company ICL bought Singer’s computing business in 1976.)

阅读大量的笔录，人们会感到整个审判过程充满了悲剧和不真实感。法官大卫·埃德尔斯坦经常被专家证人每天滔滔不绝的术语弄得不知所措；这通常导致他在下午三点左右发脾气。（法庭的空调系统有故障，夏天也无济于事。）花在聘请和留住一流律师团队（由 IBM 的尼古拉斯·卡森巴赫领导）及其研究人员上的钱并没有用于研究和开发新的计算机技术。然而，双方都有高薪的法律和研究人员，却完全错过了每个人都认识到的计算发展的明显方向：向基于微处理器的网络化桌面计算发展。没有任何记录显示有人将 Apple II 带入曼哈顿下城的法院大楼；如果有人带了，有人会认出它是什么吗？巧合的是，就在 1977 年加州计算机博览会上推出 Apple II 时，一位专家证人作证说：“我会比那更强硬一点……通用计算机行业不太可能出现任何重大的新冒险。” ²¹直到 1986 年，一位司法部经济学家仍对案件的驳回感到愤怒，他抱怨说：“IBM 面临的国内或国外竞争并不严重，不会威胁到其主导地位。” ²²这句话是在微软向公众发售股票的那一年发表的。几年后，IBM 开始遭受前所未有的损失，并开始有史以来第一次裁员。一批新书很快问世，讲述了 IBM 如何被比尔盖茨击败的故事。除了写关于 IBM 的揭秘书籍外，计算机行业的其他一切都发生了根本性的、不可逆转的变化。²³

Reading through the volumes of transcripts, one feels a sense of tragedy and unreality about the whole trial. The judge, David Edelstein, was often baffled by the avalanche of jargon that spewed forth from the expert witnesses each day; this typically resulted in his losing his temper by mid-afternoon. (The courtroom had a defective air-conditioning system, which did not help matters in the summer, either.) The money spent on hiring and retaining a team of top-notch attorneys (led by Nicholas Katzenbach for IBM) and their research staffs was money that did not go into the research and development of new computer technology. And yet both sides, with all their highly paid legal and research staffs, utterly and completely missed what everyone has since recognized as the obvious way that computing would evolve: toward microprocessor-based, networked desktop computing. There is no record of someone bringing an Apple II into the courthouse building in lower Manhattan; if someone had, would anyone have recognized it for what it was? By coincidence, just as the Apple II was being introduced at a computer fair in California in 1977, one expert witness testified, “I will be a little stronger than that ... it is most unlikely that any major new venture into the general purpose [sic] computer industry can be expected.”²¹ As late as 1986 one Justice Department economist, still fuming over the dismissal of the case, complained that “IBM faces no significant domestic or foreign competition that could threaten its dominance.”²² That statement was made the year that Microsoft offered its shares to the public. A few years later IBM began suffering unprecedented losses and began laying off employees for the first time in its history. A new crop of books soon appeared, these telling the story of how IBM had been outsmarted by Bill Gates. Other than writing tell-all books about IBM, everything else about the computer industry had fundamentally and irrevocably changed.²³

最终，争斗者精疲力竭。1981 年，罗纳德·里根的就职典礼足以结束这场争斗，他曾反对过度行使联邦权力。但真正让政府的诉讼失败的是，即使忽略个人电脑，十年间也存在着激烈而健康的竞争。通用电气和 RCA 的失败被数字设备公司、SDS、阿姆达尔和 EDS 等软件公司的成功所抵消。这个行业太健康了：吉恩·阿姆达尔和马克斯·帕列夫斯基的个人财富使这些指控很难被认真对待。24^在一次罕见的轻松情况下，IBM 的律师在描述帕列夫斯基——一名哲学系学生——在几年内用针对 IBM 中端大型机的机器赚取的巨额财富时，在法庭上引起了不少笑声。

In the end the combatants ran out of energy. The 1981 inauguration of Ronald Reagan, who had campaigned against an excessive exercise of federal power, was enough to end it. But what really killed the government’s case was that, even neglecting the personal computer, there was vigorous and healthy competition throughout the decade. The failures of GE and RCA were more than offset by the successes of Digital Equipment, SDS, Amdahl, and software companies like EDS. The industry was too healthy: the personal fortunes amassed by Gene Amdahl and Max Palevsky made it hard to take the charges seriously.²⁴ In one of the rare instances of levity, IBM’s lawyers were able to elicit more than a few chuckles in the courtroom when they described the enormous wealth that Palevsky—a philosophy student—made in a few years with machines aimed right at IBM’s middle range of mainframes.

IBM 继续开发新产品。除了 4300 和 3030 大型机外，IBM 还于 1978 年推出了 System/38，向小型机公司发起了挑战，随后又于 1988 年推出了 AS/400。AS/400 的目标客户更多是企业而非工程客户，但除此之外，它是 VAX 的强大竞争对手。它使用了 IBM 计划用于 System/370 后续产品的高级架构功能，但尚未实现。因此，AS/400 代表了 IBM 最先进的技术，并在 20 世纪 90 年代为 IBM 带来了可观的收入，当时 IBM 的大型机销售受到技术过时的影响。25 ^IBM当时未能将其他产品推向市场，这一失败最终损害了公司。目前尚不清楚反垄断诉讼与此有多大关系。

IBM continued to develop new products. In addition to the 4300 and 3030 mainframes, IBM went after the minicomputer companies with its System/38 in 1978, following that with its AS/400 in 1988. The AS/400 was aimed more at business than engineering customers, but otherwise it was a strong competitor to the VAX. It used advanced architectural features that IBM had planned for its follow-on to the System/370 but had not implemented. As such, the AS/400 represented IBM’s most advanced technology, and it generated strong revenues for IBM into the 1990s, when its mainframe sales suffered from technological obsolescence. ²⁵ IBM failed to bring other products to market at this time, however, a failure that ultimately hurt the company. It is not clear how much the antitrust suit had to do with that.

从“POTS”到“OLTP”

From “POTS” to “OLTP”

计算机公用事业的概念在 20 世纪 60 年代末被天真地设想为像电力公用事业一样，但在 20 世纪 70 年代，它朝着几个方向发展。通用电气通过与达特茅斯学院的合作建立了一个大型国际网络。其他公用事业使用 PDP-10 和 SDS 940 等机器提供非结构化的计算机时间。到 1975 年，TYMNET 包含一个由 26 台主机组成的网络，通过 80 个通信处理器进行路由。简单的轮辐式分时拓扑演变为多个环的网络，因此一台主机的故障不会导致系统崩溃^。26

The concept of a computer utility, naively envisioned in the late 1960s as being like the electric power utilities, evolved in several directions in the 1970s. General Electric built a large international network from its association with Dartmouth. Using machines like the PDP-10 and SDS 940, other utilities offered unstructured computer time. By 1975 TYMNET comprised a network of twenty-six host computers, routed through eighty communications processors. The simple hub-and-spoke topology of time-sharing evolved into a web of multiple rings, so that the failure of one host would not bring the system down.²⁶

与此同时，还出现了一种结构更严密、更规范的在线访问终端使用方法。这种方法专为特定应用而设计，例如处理保险索赔，其中终端处理的程序和数据类型受到限制。许多系统是私有的，但也有一些是半公开的，例如美国国家医学图书馆将其百年历史的《医学索引》在线化的努力。（到 20 世纪 70 年代末，其MEDLINE系统提供了来自全球研究图书馆的医学文献在线搜索。）这些系统更像 20 世纪 50 年代末的 SAGE 防空系统和 SABRE 航空预订系统，而不是达特茅斯学院的模型。出现了一个新的缩写词来描述它，即“OLTP”（代表“在线事务处理”），以将其与“POTS”（普通老式分时）的不太结构化的含义区分开来。因此，尽管计算机的使用不再是成批的卡片，但穿孔卡安装的一些基本结构仍然存在。

At the same time, a more tightly structured and disciplined use of terminals for on-line access also appeared. This was tailored for a specific application, such as processing insurance claims, where the programs and types of data handled by a terminal were restricted. Many were private, though some were semipublic, such as the effort by the U.S. National Library of Medicine to put its century-old Index Medicus on-line. (By the end of the 1970s its MEDLINE system provided on-line searches of medical literature from research libraries worldwide.) These systems were more like the SAGE air-defense and SABRE airline reservations systems of the late 1950s than they were like the Dartmouth College model. A new acronym appeared to describe it, “OLTP” for “On-line Transaction Processing,” to differentiate it from the less-structured connotations of “POTS” (Plain Old Time-Sharing). Thus although computer usage was no longer in batches of cards, some of the basic structure of a punched-card installation remained.

许多公司推出了终端来服务这个市场。有些终端是从 Teletype 衍生而来，结合了打字机键盘和打印机（例如 DECwriter II 或 Teletype Model 37，均于 1975 年左右推出）。其他公司复制了 Teletype，只是用视频屏幕代替了打印机。这些终端，如 Lear-Siegler ADM-3，有时被称为“哑终端”、“玻璃电传打字机”或“玻璃 TTY”：它们除了简单的数据输入和查看外，几乎没有其他功能。与它们形成对比的是“智能”终端，它允许用户查看和编辑全屏文本，并具有一定的计算能力。除了 VT-100 之外，DEC 还生产了几款围绕 PDP-8 处理器设计的终端；另一家取得一定成功的公司是位于圣安东尼奥的 Data-point。回想一下，正是 Datapoint 与英特尔的合同促成了 8080 微处理器的诞生；然而，Datapoint 2200 终端并未使用微处理器。其中一些终端，尤其是 Datapoint，几乎成为了个人电脑，而供应商却没有意识到这一点^。27

A number of companies introduced terminals to serve this market. Some were descended from the Teletype and combined a typewriter keyboard and a printing mechanism (e.g., the DECwriter II or Teletype Model 37, both ca. 1975). Others replicated the Teletype, only with a video screen instead of a printer. These, like the Lear-Siegler ADM-3, were sometimes called “dumb terminals,” “glass teletypes,” or “glass TTY”: they offered little beyond simple data entry and viewing. In contrast to them were “smart” terminals that allowed users to see and edit a full screen of text, and which contained a measure of computing power. Besides the VT-100, DEC had produced several designed around a PDP-8 processor; another company that had some success was Data-point of San Antonio. Recall that it was Datapoint’s contract with Intel that led to the 8080 microprocessor; however, the Datapoint 2200 terminal did not use a microprocessor. Some of these terminals, especially the Datapoint, came close to becoming personal computers without the vendor realizing it.²⁷

VT-100 成为标准 ASCII 终端，而 IBM 推出的终端在 1980 年成为 EBCDIC 标准。这就是 1971 年发布的 3270 型号。28 ³²⁷⁰在理念上与 DEC VT-100 相反：它假设用户将结构化信息键入选定字段，而其他字段（例如，用于个人姓名或生日）则会为每个记录重复输入。因此，终端在键入信息时不会传输信息，而是等到键入全屏信息后才发送；然后它只发送计算机的新内容（以压缩形式）。IBM 大型机安装现在通常包括终端和通过分时选项 (TSO) 软件实现的分时访问。通常这些终端被隔离在大型机安装附近的特殊房间中。它们很少出现在私人办公室中。

The VT-100 became the standard ASCII terminal, while a terminal introduced by IBM became the EBCDIC standard by 1980. That was the model 3270, announced in 1971.²⁸ The 3270 was the philosophical opposite of the DEC VT-100: it operated on the assumption that the user would be keying structured information into selected fields, with other fields (e.g., for a person’s name or date of birth) replicated over and over for each record. Therefore, the terminal did not transmit information as it was keyed in but waited until a full screen was keyed in; then it sent only whatever was new to the computer (in compressed form). IBM mainframe installations now routinely included terminals and time-shared access through the time sharing option (TSO) software. Typically these terminals were segregated in special rooms near the mainframe installation. They were seldom found in a private office.

到 1980 年，随着诉讼即将结束，IBM 仍然主导着行业。但 IBM 越来越处于较慢的河道中。这始于 1963 年 ASCII 的开发，当时 IBM 采用了 EBCDIC。1964 年，IBM 选择了混合半导体技术而非 IC。1970 年，它采用了自己设计的集成电路，与当时充斥市场的标准 TTL 芯片略有不同。70 年代中期，IBM 的系统网络架构为大型系统的联网建立了标准，但 SNA 与当时国防部正在开发的联网方案不同。最后，3270 和 VT-100 代表了不同的终端设计方法。只有事后我们才能看出其中的规律。

By 1980, as the lawsuit was coming to an end, IBM still dominated the industry. But more and more, IBM was floating in a slower channel of the river. That began in 1963 with the development of ASCII, when IBM adopted EBCDIC. In 1964 IBM chose a hybrid semiconductor technology over ICs. In 1970 it adopted integrated circuits of its own design, slightly different from the standard TTL chips then flooding the market. In the mid-70s, IBM’s Systems Network Architecture established a standard for networking large systems, but SNA was different from the networking schemes then being developed by the Defense Department. Finally, there were the different approaches to terminal design represented by the 3270 and VT-100. Only with hindsight can we discern a pattern.

IBM 于 1981 年推出个人电脑，使问题达到了顶峰。IBM PC 使用 ASCII，而不是 EBCDIC。它使用来自外部供应商的标准 TTL 和 MOS 芯片。而且它与键盘和显示器的连接更接近小型计算机，而不是 3270。PC 的内部设计揭示了市场压力如何能够完成法院和美国司法部无法完成的事情。

IBM’s introduction of the personal computer in 1981 brought the issue to a head. The IBM PC used ASCII, not EBCDIC. It used standard TTL and MOS chips from outside suppliers. And its connections with its keyboard and monitor were closer to the minicomputer than to the 3270. The PC’s internal design reveals how the pressures of the marketplace were able to accomplish what the courts and the U.S. Justice Department could not do.

大型机、批量计算模型（其巅峰是 7090）正在让位于交互式处理和分散式处理。区分“智能”终端和独立个人计算机的界限越来越模糊就是一个迹象。新的设计问题浮出水面：如何在终端和中央系统之间分配处理功能和内存，以及如何通过这样的网络高效可靠地发送数据。IBM 已着手设计“未来系统”（FS），以正面解决其中一些问题。FS 的规划始于 20 世纪 70 年代初，IBM 希望在 1975 年前推出替代其 System/370 系列的产品。但 FS 在 1975 年被放弃，部分原因是其设计人员无法解决架构问题，部分原因是 System/370 架构的成功意味着 IBM 将面临不可接受的风险，将该市场拱手让给第三方供应商。²⁹一些概念被运用到了中端 AS/400 中，但取消 FS 是“IBM 历史上代价最昂贵的开发失败” ^。30

The mainframe, batch model of computing, whose high-water mark was the 7090, was giving way, not only to interactive but also to decentralized processing. The increasingly fuzzy line that distinguished “smart” terminals from stand-alone personal computers was one indication. New design questions came to the fore: how to apportion processing functions and memory among the terminals and a central system, and how to send data efficiently and reliably through such a network. IBM had embarked on the design of a “Future System” (FS) that attacked some of these issues head-on. Planning for FS began in the early 1970s, and IBM hoped to announce products to replace its System/ 370 line by 1975. But FS was abandoned in 1975, in part because its designers were unable to solve the architectural problems, and in part because the success of the System/370 architecture meant that IBM would put itself at an unacceptable risk to abandon that market to third-party vendors.²⁹ Some of the concepts found their way into the midrange AS/400, but canceling FS was “the most expensive development-effort failure in IBM’s history.”³⁰

维亚特龙

Viatron

128 号公路上的一家初创公司也曾有过类似的设想，但结果同样令人沮丧。这家公司就是位于马萨诸塞州贝德福德的 Viatron 计算机系统公司，它堪称分布式计算领域的施洗约翰。它是 20 世纪 60 年代中期空军的一项名为 AESOP（高级在线规划实验系统）的项目的产物。AESOP 由 MITRE 公司筹备，设想了一个终端网络，为中高层管理人员（包括那些对计算技术一无所知的管理人员）提供视觉和文本信息，帮助他们以与电话相同的接受度完成工作：

A start-up company from Route 128 had an idea with similar promise but equally dismal results. The John the Baptist of distributed computing was Viatron Computer Systems of Bedford, Massachusetts. It was the outgrowth of an Air Force Project from the mid-1960s called AESOP (Advanced Experimental System for On-line Planning). Prepared by the MITRE Corporation, AESOP envisioned a network of terminals that provided visual as well as text information to middle and high-level managers, including those without any sophistication in computing, to help them do their work with the same level of acceptance as the telephone:

管理系统的核心……将不是中央处理器或中央存储器。真正的基础……将是独特的指令程序，它使中央处理器、中央存储器以及组织的数据存储和正式的定量模型可以通过其桌面显示器的窗口轻松供管理人员使用，从而使他能够通过使用简单的光枪指针充分发挥其意图的力量。随着 AESOP 类型管理系统的发展，管理人员将学会轻松自然地与处理器交谈和交互。他们还将学会通过处理器与组织的其他成员进行交流^。31

The core of the management system ... will be not so much the central processor or central memory. The real basis ... will be the unique program of instructions which makes the central processor, the central memory, and the organization’s store of data and formal quantitative models easily available to the manager through the window of his desk top display, thus making it possible for him to exert the full power of his intentions through the use of his simple lightgun pointer. As AESOP-type management systems are developed, managers will learn to converse and interact with the processor with ease and naturalness. They will also learn to communicate through the processor with other members of the organization.³¹

报告的两位作者约瑟夫·斯皮格尔和爱德华·贝内特博士于 1967 年离开 MITRE，共同创办了 Viatron。贝内特成功筹集了风险投资——当时正是黄金时代——并宣布，到 1969 年，Viatron 将出租交互式终端，将处理转移到桌面上。他还预测，他的公司将在安装的计算机数量上超过 IBM。System 21 终端的租金低得令人难以置信，每月仅 40 美元。32^该系统包括一个键盘、一个 9 英寸视频显示器和两个用于存储数据和格式化信息的盒式磁带驱动器（图 8.2）。一个可选附件允许用户断开键盘和磁带机的连接，并将其连接到任何标准电视机上，以便在酒店房间等场合进行远程计算。终端内部包含一个可存储 512 个字符的“微处理器”。其他选项包括光学字符识别设备、“通信适配器”以及一个巧妙的、受鲁布·戈德堡启发的“打印机器人”，该机器人放置在标准 IBM Selectric 打字机上。机械手指启动一组螺线管，按下 Selectric 键，以每秒 12 个字符的速度输入干净的输出^。33

Two of the report’s authors, Joseph Spiegel and Dr. Edward Bennett, left MITRE and cofounded Viatron in 1967. Bennett was successful in raising venture capital—these were the go-go years—and announced that by 1969 Viatron would be renting interactive terminals that would move processing onto the desktop. He also predicted that his company would surpass IBM in numbers of installed computers. System 21 terminals were to rent for the unbelievably low price of $40 a month.³² The system included a keyboard, a 9-inch video display, and two cassette tape drives for storage of data and formatting information (figure 8.2). An optional attachment allowed users to disconnect the keyboard and tape unit and connect it to any standard television set for remote computing, say, in a hotel room. The terminal contained within it a “micro-processor” [sic] with 512 characters of memory. Other options included an optical character-recognition device, a “communications adapter,” and an ingenious, Rube Goldberg–inspired “printing robot” which one placed over a standard IBM Selectric typewriter. Activating a set of solenoids, mechanical fingers pressed the Selectric keys to type clean output at 12 characters/second.³³

Viatron 令人印象深刻的规格的关键在于它使用了 MOS 集成电路。这种集成电路制造技术是 20 世纪 70 年代微处理器革命的技术基础，但在 1969 年还不成熟。Viatron 必须将其启动资金投入到完善 MOS 中，然后它需要更多资金来为批量生产做好准备。这太过雄心勃勃了。到 1970 年，生产线才刚刚起步，但产量很小，Viatron 的销售和营销陷入混乱。1970 年夏天，在 Bennett 家举行的董事会会议上，他发现自己被赶出了 Viatron，而此时他的妻子正要为每个人准备晚餐（他们从未吃过这顿饭）。到 1971 年，该公司交付了几套系统，但在同年 4 月宣布破产。仅在 1969-1970 财年，风险投资家的损失就高达 3000 万美元。³⁴ Viatron 只是众多试图取代 IBM 成为行业领头羊但却失败的公司之一。

The key to Viatron’s impressive specifications was its use of MOS integrated circuits. This technique of integrated-circuit fabrication was the technical foundation for the microprocessor revolution of the 1970s, but it was immature in 1969. Viatron had to invest its start-up capital in perfecting MOS, and then it needed more money to gear up for volume production. That was too ambitious. By 1970, production lines were just starting, but the volume was small, and Viatron’s sales and marketing were in disarray. At a meeting of the board held in Bennett’s home in the summer of 1970, he found himself ousted from Viatron just as his wife was about to serve everyone dinner (they never ate the meal). The company delivered a few systems by 1971, but in April of that year it declared bankruptcy. Losses to venture capitalists ran upwards of $30 million in fiscal year 1969–1970 alone.³⁴ Viatron became just another of many companies to fail while attempting to topple IBM from the top of the industry.

图 8.2

办公自动化：Viatron 21。（来源：明尼苏达大学查尔斯巴贝奇研究所）

Figure 8.2

Office automation: Viatron 21. (Source: Charles Babbage Institute, University of Minnesota.)

王

Wang

即使以当时的繁荣时期来衡量，Viatron 的发展轨迹也显得怪异，但这不应掩盖 Bennett 观察的真相。MOS 集成电路的进步使 IBM 的计算方式过时，至少在办公环境中是如此。在 Viatron 失败的地方取得成功的公司是王安实验室，该公司在早期曾是电子计算器领域的先驱。到 1971 年，王安实验室意识到计算器正在成为一种商品，其微薄的利润率更多地取决于包装而非技术创新。王安实验室开始向小型计算机公司转型，并于 1972 年通过其 Model 2200“计算计算器”实现了完全跨界——这是一款通用计算机，尽管王安实验室小心谨慎，没有将其作为通用计算机进行营销。与 20 世纪 50 年代末的数字设备公司一样，王安实验室也不愿意使用“计算机”一词，因为该词的含义太深。王先生敏锐地知道何时退出一个市场，进入一个即将开放的新市场。随着王先生的利润飙升，华尔街分析师们编造出各种复杂的理论来解释这一点，其中一些理论是基于他的中国祖先赋予他的一种天生的感觉。事实上，王博士是一位保守的工程师，他了解公司产品的技术，并且重视公司的独立性。在这方面，他更接近他的美国同行兼邻居、DEC 公司的肯·奥尔森，而不是西方人对东方思维模式的任何刻板印象^。35

Even by the metric of the go-go years, Viatron’s trajectory was bizarre, which should not obscure the truth of Bennett’s observation. Advances in MOS integrated circuits were making IBM’s way of doing computing obsolete, at least in the office environment. The company that succeeded where Viatron failed was Wang Labs, which in an earlier era had pioneered in electronic calculators. By 1971 Wang recognized that calculators were becoming a commodity, with razor-thin profit margins dependent on packaging more than on technical innovation. Wang Labs began a transition to a minicomputer company, and by 1972 made a complete crossover with its Model 2200 “computing calculator”—a general-purpose computer, although Wang was careful not to market it as such. Like Digital Equipment in the late 1950s, Wang was reluctant to use the word “computer” because of the word’s connotations. Wang had an astute sense of knowing when to get out of one market and into a new one about to open up. As Wang’s profits soared, Wall Street analysts concocted elaborate theories to explain this, some based on an alleged innate sense that his Chinese ancestors gave him. Dr. Wang was, in fact, a conservative engineer who understood the technology of his company’s products and who valued his company’s independence. In this regard he was closer to his Yankee counterpart and neighbor, Ken Olsen of DEC, than he was to any Western stereotype of Oriental mind-set.³⁵

王安接下来选择将公司引向后来被称为“办公自动化”的领域。在 20 世纪 70 年代中期，这意味着文字处理。文字处理已经变得如此普遍，以至于很难回忆起在当时这个概念有多么荒谬，当时即使是小型计算机也要花费数千美元，而熟练的打字员每小时的工资只有 1.25 美元。一个古老的故事讲述了麻省理工学院的研究生如何在 20 世纪 60 年代初对价值 12 万美元的 PDP-1 进行编程，以用作“昂贵的打字机”。IBM 为 7090 开发了一个名为 TEXT-90 的程序，但该程序仅用于特殊应用，从未进入办公环境。在 20 世纪 60 年代初，一些从事 ASCII 标准工作的委员会成员认为小写字母的代码是不必要的，而且浪费空间^。36

An Wang chose next to direct the company toward what would later be known as “Office Automation.” In the mid-1970s that meant word processing. Word processing has become so commonplace that it is hard to recall how absurd the concept was at a time when even small computers cost thousands and skilled typists were paid $1.25 an hour. An old story tells of how graduate students at MIT programmed the $120,000 PDP-1 to serve as an “expensive typewriter” in the early 1960s. IBM developed a program called TEXT-90 for the 7090, but that was used only for special applications and never penetrated the office environment. In the early 1960s, some members of the committee working on the ASCII standard argued that codes for lowercase letters were unnecessary and a waste of space.³⁶

“文字处理”一词开始使用是在 1964 年之后，当时 IBM 发布了 MTST，这是 Selectric 打字机的一个版本，可以在磁带盒上存储和调用按键序列。37^王安公司的早期产品 Model 1200 与之类似，但客户发现它的复杂性令人望而生畏。当时进入该领域的其他公司包括丹佛郊外的 NBI（“Nothing But Initials”）、亚特兰大的 Lanier 和明尼阿波利斯的 CPT。

The term “word processing” came into use after 1964, when IBM announced the MTST—a version of its Selectric typewriter that could store and recall sequences of keystrokes on a magnetic tape cartridge.³⁷ An early Wang product, the Model 1200, was similar, but customers found its complexity daunting. Other companies that entered the field at this time included NBI (“Nothing But Initials”) outside of Denver, Lanier in Atlanta, and CPT in Minneapolis.

第二次，王安公司就成功了。王安公司的工程师首先发现了办公室人员的需求。他们意识到，许多文字处理设备的用户都害怕因为不小心按错键而丢失一天的工作成果。而且，不仅仅是秘书容易犯这样的错误：1981 年，前总统吉米·卡特在价值 12,000 美元的兰尼尔“没问题”文字处理系统上按错键，丢失了几页回忆录——“我花了好几天才写好”。他焦急地给兰尼尔打了个电话，兰尼尔拿出了一张实用磁盘，让他能够从原始磁盘中恢复数据。38^此后，王安公司的工程师想出了一个设计，使这种丢失几乎不可能发生。他们还决定使用阴极射线管终端，它可以显示半页文本，而不是其他系统使用的一行或几行。命令通过一个简单的菜单屏幕访问。在后来的时代，王安电脑的设计可能会以“用户友好”这个陈词滥调而闻名；它也是一个“分布式”系统。但该公司在营销中没有使用这两个术语。与其他小型计算机公司不同，王安电脑很少做 OEM 业务；它把机器卖给将要使用它的人。王安电脑的客户——最初是华尔街经纪公司、大型银行和石油公司——不用技术术语。（十年后，办公室工作人员就没那么幸运了，无论“用户友好”这个词多么晦涩难懂，所有东西都会贴上标签。）³⁹

The second time around Wang got it right. Wang engineers found out first of all what office people wanted. They realized that many users of word-processing equipment were terrified of losing a day’s work by the inadvertent pressing of the wrong key. And it wasn’t just secretaries who were prone to such actions: in 1981 ex-President Jimmy Carter lost a few pages of his memoirs—”I had labored over them for a couple of days”— by pressing the wrong key on his $12,000 Lanier “No Problem” [sic] word-processing system. An anxious phone call to Lanier produced a utilities disk that allowed him to recover the data from the original diskette.³⁸ After this, Wang’s engineers came up with a design that would make such a loss nearly impossible. They also decided on a terminal that used a cathode ray tube, which displayed half a page of text instead of the one or few lines that other systems used. Commands were accessed by a simple screen of menus. In a later era Wang’s design might have been known by the cliche´ “user-friendly”; it was also a “distributed” system. But the company used neither term in its marketing. Unlike other minicomputer companies, Wang did little OEM business; it sold machines to the people who were going to use it. Wang spared its customers—Wall Street brokerage houses, large banks, and oil companies at first—the technical jargon. (A decade later office workers were not so lucky, everything would get plastered with the term “user friendly” no matter how obtuse it was.)³⁹

该系统的主要要求是快速响应。分时系统让用户无需拿着一叠打孔卡排队等候，但在繁忙的日子里，用户同样需要在他们的终端前等待，等待主机处理每项工作。与麻省理工学院的黑客不同，办公室职员不可能半夜来上班。解决办法是将部分处理能力放在终端本身，中央计算机主要用于数据存储和检索——这在 1985 年之后很常见，但在 1975 年却与分时系统截然不同。WPS（王氏文字处理系统）于 1976 年 6 月在纽约的一次贸易展上亮相，据一些报道，该系统几乎引发了一场骚乱（图 8.3）。40^包括硬盘存储在内的基本系统售价为 30,000 美元。王安实验室在 1976 年的数据处理收入排名中位列第 45 位，到 1983 年上升到第 8 位，仅次于 IBM、DEC 和 BUNCH 的残余。一些分析师认为王安公司最有可能成为行业第二名。（没有华尔街人士会冒着失去职业生涯的风险预测新的第一名。）其他人将公司的成功归类为“办公自动化”而不是通用计算，但王安公司销售的其实是通用分布式计算机系统。王安公司的成功证明了 Viatron 的愿景是正确的。然而，王安公司在 20 世纪 90 年代无法再次重塑自我，当时它面临着运行廉价文字处理软件的商品个人电脑的竞争，最终也破产了^。41

A major requirement was that the system have a speedy response. Time-sharing relieved users of the need to wait in a queue with a deck of punched cards, but on a busy day users faced an equally onerous wait at their terminals while the mainframe got around to each job. Unlike MIT hackers, office employees could not be expected to come in at midnight to do their work. The answer was to put some of the processing power into the terminal itself, with the central computer serving primarily for data storage and retrieval—commonplace after 1985, but a radical departure from time-sharing in 1975. The WPS (Wang Word Processing System) was unveiled at a trade show in New York in June 1976, and according to some accounts nearly caused a riot (figure 8.3).⁴⁰ A basic system, including hard disk storage, cost $30,000. Wang Labs, ranked forty-fifth in data-processing revenues in 1976, moved up to eighth place by 1983, just below IBM, DEC, and the remnants of the BUNCH. Some analysts thought Wang was in the best position of any company to become number two in the industry. (No Wall Street person would risk his career by predicting a new number one.) Others put the company’s success into the pigeonhole of “office automation” rather than general-purpose computing, but what Wang was selling was at heart a general-purpose, distributed computer system. Wang’s success was a vindication of Viatron’s vision. However, Wang was unable to reinvent itself once again in the 1990s, when it faced competition from commodity personal computers running cheap word-processing software, and it too went bankrupt.⁴¹

图 8.3

办公自动化：WANG 文字处理系统。（来源：明尼苏达大学查尔斯·巴贝奇研究所。）

Figure 8.3

Office automation: WANG Word Processing System. (Source: Charles Babbage Institute, University of Minnesota.)

施乐帕克研究中心

Xerox PARC

王安公司故事的讽刺之处在于，尽管王安公司有诸多创新，但 20 世纪 70 年代的报道却很少提及王安公司。阅读这些主题的文献，人们会得出这样的结论：施乐公司才是分布式、用户友好型计算领域的真正先驱；施乐帕洛阿尔托研究中心是计算机未来诞生的地方，斯图尔特·布兰德在 1972 年的《滚石》杂志文章中对此大加赞赏。为什么会这样呢？

One of the ironies of the story of Wang is that despite its innovations, few stories written about the 1970s talk about Wang. To read the literature on these subjects, one would conclude that the Xerox Corporation was the true pioneer in distributed, user-friendly computing; that the Xerox Palo Alto Research Center, which Stewart Brand so glowingly described in his 1972 Rolling Stone article, was the place where the future of computing was invented. Why was that so?

1970 年，施乐公司在帕洛阿尔托山麓建立了一个研究实验室。其目标是预测技术将给商业世界的信息处理方式带来的深刻变化。作为一家以复印机闻名的公司，施乐对“无纸化办公室”的言论感到紧张是可以理解的。施乐不知道这是否真的会发生，但它希望其帕洛阿尔托研究中心 (PARC) 能帮助公司渡过难关^。42

The Xerox Corporation set up a research laboratory in the Palo Alto foothills in 1970. Its goal was to anticipate the profound changes that technology would bring to the handling of information in the business world. As a company famous for its copiers, Xerox was understandably nervous about talk of a “paperless office.” Xerox did not know if that would in fact happen, but it hoped that its Palo Alto Research Center (PARC) would help the company prosper through the storms.⁴²

有两件事让 PARC 的成立对计算机行业具有重大意义。首先是选择帕洛阿尔托：施乐公司企业研究总监雅各布·戈德曼 (Jacob Goldman) 原本偏爱康涅狄格州纽黑文，但他聘请来建立实验室的乔治·帕克 (George Pake) 却偏爱帕洛阿尔托，并最终胜出，尽管那里距离施乐公司纽约州北部的运营基地和康涅狄格州总部很远。该实验室成立时，由新成立的英特尔公司的罗伯特·诺伊斯 (Robert Noyce) 领导的“硅谷”刚刚成型。

Two things made PARC’s founding significant for computing. The first was the choice of Palo Alto: Jacob Goldman, director of corporate research at Xerox, had favored New Haven, Connecticut, but the person he hired to set up the lab, George Pake, favored Palo Alto and prevailed, even though it was far from Xerox’s upstate New York base of operations and its Connecticut headquarters. The lab opened just as “Silicon Valley,” led by Robert Noyce of the newly founded Intel, was taking form.

PARC 具有重要意义的第二个原因发生在国会大厅。随着大学校园内对美国参与越南战争的抗议声不断高涨，国会中也展开了一场平行辩论，其中包括大学作为国防部资助战争相关研究的场所的角色。参议员 J. William Fulbright 特别批评了这种观点，他认为科学研究在“军事工业综合体”（艾森豪威尔总统于 1961 年创造的术语）的“巨无霸”面前正在失去独立性。在 1970 年《军事采购授权法案》的修正案中，由参议员 Mike Mansfield 主持的一个委员会插入了以下措辞：“任何授权资金……不得用于开展任何研究项目或研究，除非此类研究与特定军事职能或行动有直接和明显的关系。” ⁴³该委员会并非旨在削弱大学的基础研究，只是要将基础研究与应用研究分开。一些成员认为，国家科学基金会将取代国防部资助基础研究。早在《曼斯菲尔德修正案》通过之前，美国国防部就已开始减少与特定武器系统无关的研究经费，因此这一运动得到了鹰派和鸽派的支持。

The second reason for PARC’s significance took place in the halls of Congress. As protests mounted on college campuses over the U.S. involvement in Viet Nam, a parallel debate raged in Congress that included the role of universities as places where war-related research was being funded by the Defense Department. Senator J. William Fulbright was especially critical of the way he felt science research was losing its independence in the face of the “monolith” of the “militaryindustrial complex” (a term coined by President Eisenhower in 1961). In an amendment to the 1970 Military Procurement Authorization Bill, a committee chaired by Senator Mike Mansfield inserted language that “none of the funds authorized ... may be used to carry out any research project or study unless such a study has a direct and apparent relationship to a specific military function or operation.”⁴³ The committee did not intend to cripple basic research at universities, only to separate basic from applied research. Some members assumed that the National Science Foundation would take the DoD’s place in funding basic research. Even before the passage of this “Mansfield Amendment,” the DoD had moved to reduce spending on research not related to specific weapons systems; thus this movement had support among hawks as well as doves.

NSF 从未获得足够的资源来弥补这一空缺。在几所精选的大学里，那些从事高级计算基础研究的人感到他们处于危险之中，因为他们的工作几乎完全由国防部的高级研究计划局 (ARPA) 资助。44^就在那时，乔治·帕克正在全国的大学里寻找人员来为施乐 PARC 工作。他找到了一批有才华和雄心勃勃的人愿意搬到帕洛阿尔托。ARPA 的资金并非不分青红皂白，而是主要集中在几所大学——麻省理工学院、卡内基梅隆大学、斯坦福大学、加州大学伯克利分校、加州大学洛杉矶分校和犹他大学——几乎所有大学的研究人员最终都进入了 PARC，包括犹他州的 Alan Kay 和 Robert Taylor，以及麻省理工学院的 Jerome Elkind 和 Robert Metcalfe。⁴⁵还有一些关键人物从其他公司调任，特别是伯克利计算机公司 (BCC)，这是一家陷入困境的分时公司，是 ARPA 资助的改造 SDS 计算机以进行分时的项目的产物。查克·塔克 (Chuck Thacker) 和巴特勒·兰普森 (Butler Lampson) 是调到 PARC 的伯克利计算机校友之一。上述所有人在其职业生涯的某个阶段都曾获得 ARPA 的资助，泰勒曾担任 ARPA 信息处理技术办公室的负责人。

The NSF was never given the resources to take up the slack. At a few select universities, those doing advanced basic research on computing felt that they were at risk, because their work was almost entirely funded by the Defense Department’s Advanced Research Projects Agency (ARPA).⁴⁴ At that precise moment, George Pake was scouring the country’s universities for people to staff Xerox PARC. He found a crop of talented and ambitious people willing to move to Palo Alto. ARPA funding had not been indiscriminate but was heavily concentrated at a few universities—MIT, Carnegie-Mellon, Stanford, UC-Berkeley, UCLA, and the University of Utah—and researchers from nearly every one of them ended up at PARC, including Alan Kay and Robert Taylor from Utah, and Jerome Elkind and Robert Metcalfe from MIT.⁴⁵ There were also key transfers from other corporations, in particular from the Berkeley Computer Corporation (BCC), a struggling time-sharing company that was an outgrowth of an ARPA-funded project to adapt an SDS computer for time-sharing. Chuck Thacker and Butler Lampson were among the Berkeley Computer alumni who moved to PARC. All those cited above had had ARPA funding at some point in their careers, and Taylor had been head of ARPA’s Information Processing Techniques Office.

两位未转入 PARC 的 ARPA 研究人员启发了施乐新实验室的发展。他们是心理学家 JCR Licklider，他于 1962 年开始推动 ARPA 进军高级计算机研究；还有一位电气工程师 Douglas Engelbart，他曾在斯坦福研究所工作，后来转入 Tymshare。1960 年，在剑桥公司 Bolt Beranek and Newman 工作期间，Licklider 发表了一篇题为“人机共生”的论文，他在论文中预测了计算的未来，“将涉及人类和电子成员之间的紧密结合”。在随后的论文“计算机作为通信设备”中，他详细阐述了他的计划。46^他是在批处理鼎盛时期写作的，但在他的论文中，Licklider 指出了几个他认为可以克服的技术障碍。其中一些涉及硬件限制，而计算机电路的现有趋势很快就会克服这些限制。他认为开发高效的分时操作至关重要。其他障碍则更加难以克服：重新定义当时的编程和数据存储概念。1962 年，“Lick”加入 ARPA，在那里他被授予了一笔基金的控制权，他可以用这笔基金来实现创造“机械延伸的人”的愿景^。47

Two ARPA researchers who did not move to PARC were the inspiration for what would transpire at Xerox’s new lab. They were J.C.R. Licklider, a psychologist who initiated ARPA’s foray into advanced computer research beginning in 1962, and Douglas Engelbart, an electrical engineer who had been at the Stanford Research Institute and then moved to Tymshare. In 1960, while employed at the Cambridge firm Bolt Beranek and Newman, Licklider published a paper titled “Man-Computer Symbiosis” in which he forecast a future of computing that “will involve a very close coupling between the human and electronic members of the partnership.” In a following paper, “The Computer as a Communication Device,” he spelled out his plan in detail.⁴⁶ He was writing at the heyday of batch processing, but in his paper Licklider identified several technical hurdles that he felt would be overcome. Some involved hardware limits, which existing trends in computer circuits would soon overcome. He argued that it was critical to develop efficient time-sharing operations. Other hurdles were more refractory: redefining the notions of programming and data storage as they were then practiced. In 1962 “Lick” joined ARPA, where he was given control over a fund that he could use to realize this vision of creating a “mechanically extended man.”⁴⁷

道格拉斯·恩格尔巴特是 1962 年底第一批向 ARPA 信息处理技术办公室申请资助的人之一；他当时正在寻求对“增强人类智力”的“概念框架”的支持。48^{恩格尔巴特}说，一次偶然的机会，他看到了范尼瓦尔·布什在《大西洋月刊》上发表的文章《正如我们所想》（发表于 1945 年 7 月），这激发了他制定这样一个计划。利克莱德指示他通过一条专线与斯坦福大学合作，使用位于圣莫尼卡的分时 Q-32 实验计算机；后来，恩格尔巴特的团队使用了 CDC 160A，即原型小型计算机。该小组花时间研究和试验改善人机通信的方法。他最著名的发明是“鼠标”，最早于 1967 年被描述，经过详尽的测试，它比光笔（用于 SAGE）、操纵杆或其他输入设备更高效、更有效。⁴⁹恩格尔巴特回忆说，他受到一种叫做面积计的装置的启发，工程师可以把它滑过图表来计算曲线下的面积。在许多工程师中，这种小巧的装置就像计算尺一样常见；现在只有在古董商和博物馆里才能找到它。

Douglas Engelbart was one of the first persons to apply for funding from ARPA’s Information Processing Techniques Office in late 1962; he was seeking support for a “conceptual framework” for “augmenting human intellect.”⁴⁸ Engelbart says that a chance encounter with Vannevar Bush’s Atlantic Monthly article “As We May Think” (published in July 1945) inspired him to work on such a plan. Licklider directed him to work with the time-shared Q-32 experimental computer located in Santa Monica, through a leased line to Stanford; later Engelbart’s group used a CDC 160A, the proto-minicomputer. The group spent its time studying and experimenting with ways to improve communication between human beings and computers. His most famous invention, first described in 1967, was the “mouse,” which exhaustive tests showed was more efficient and effective than the light pen (used in the SAGE), the joystick, or other input devices.⁴⁹ Engelbart recalled that he was inspired by a device called a planimeter, which an engineer slid over a graph to calculate the area under a curve. Among many engineers this compact device was a common as a slide rule; it is now found only among antique dealers and museums.

1968 年 12 月，恩格尔巴特和十几名助手（其中包括斯图尔特·布兰德）在旧金山秋季联合计算机会议上雄心勃勃地展示了他的“增强知识研讨会”。通过投影视频屏幕系统和与帕洛阿尔托的计算机链接，向观众展示了由鼠标控制的交互式计算机程序。令人惊讶的是，一切都很顺利。尽管恩格尔巴特后来表示他对观众的反应感到失望，但这次演讲后来成为交互式计算史上的传奇。恩格尔巴特没有加入施乐-PARC，但他的许多同事，包括比尔英格利什（负责鼠标的细节设计）都加入了^。50

In December 1968 Engelbart and a crew of over a dozen helpers (among them Stewart Brand) staged an ambitious presentation of his “Augmented Knowledge Workshop” at the Fall Joint Computer Conference in San Francisco. Interactive computer programs, controlled by a mouse, were presented to the audience through a system of projected video screens and a computer link to Palo Alto. Amazingly, everything worked. Although Engelbart stated later that he was disappointed in the audience’s response, the presentation has since become legendary in the annals of interactive computing. Engelbart did not join Xerox-PARC, but many of his coworkers, including Bill English (who did the detail design of the mouse), did.⁵⁰

鼠标有什么特别之处？鼠标提供了一种实用而优越的与计算机交互的方法，不会对用户的符号推理能力造成压力。从机器诞生之初，人们就认识到了编程的困难。大多数人只需接受最少的指导，偶尔参考一下放在手套箱里的车主手册，就能学会如何驾驶汽车——这是一种复杂的设备，如果使用不当，可能会致命。汽车的控制系统向驾驶员提供了转动方向盘和改变方向、踩下油门和加速、踩下刹车和减速之间的明确直接联系。相比之下，UNIX 的命令只有两个或三个字母，其中删除文件的命令可能与打印文件的命令仅在相邻键上有所不同。汽车和鼠标使用眼手协调，这是人类在数千年的进化过程中学会的技能，但键盘使用的是人类最近才掌握的思维模式。 PARC 的研究人员对鼠标进行了改进，并将其集成到视频屏幕上的视觉显示和图标符号系统（人类认知的另一个未充分利用的维度）。

What was so special about the mouse? The mouse provided a practical and superior method of interacting with a computer that did not strain a user’s symbolic reasoning abilities. From the earliest days of the machine’s existence, the difficulties of programming it were recognized. Most people can learn how to drive a car—a complex device and lethal if not used properly—with only minimal instruction and infrequent reference to an owner’s manual tossed into the glove box. An automobile’s control system presents its driver with a clear, direct connection between turning the steering wheel and changing direction, pressing on the gas pedal and accelerating, pressing on the brake pedal and slowing down. Compare that to, say, UNIX, with its two-or three-letter commands, in which the command to delete a file might differ from one to print a file only by adjacent keys. Automobiles—and the mouse—use eye-hand coordination, a skill human beings have learned over thousands of years of evolution, but a keyboard uses a mode of human thought that humans acquired comparatively recently. Researchers at PARC refined the mouse and integrated it into a system of visual displays and iconic symbols (another underutilized dimension of human cognition) on a video screen.

对于美国计算机行业来说，研究从 ARPA 转移到施乐是一件好事；它将成本和营销参数强加到他们的产品上。据说施乐未能成功过渡到商业产品；正如一位作家所描述的那样，它“摸索着走向未来”。苹果，而不是施乐，在 1984 年推出了 Macintosh，将窗口、图标、鼠标和下拉菜单（WIMP 界面）的概念带入大众市场。施乐发明了一种名为以太网的网络方案，并于 1980 年将其推向市场（与 Digital 和英特尔联合推出），但像 3-Com 这样的小公司仍然在广泛商业化以太网。惠普将激光打印机商业化，这是另一项施乐-PARC 创新。等等^。51

For the U.S. computing industry, the shift of research from ARPA to Xerox was a good thing; it forced the parameters of cost and marketing onto their products. It is said that Xerox failed to make the transition to commercial products successfully; it “fumbled the future,” as one writer described it. Apple, not Xerox, brought the concept of windows, icons, a mouse, and pull-down menus (the WIMP interface) to a mass market, with its Macintosh in 1984. Xerox invented a networking scheme called Ethernet and brought it to market in 1980 (in a joint effort with Digital and Intel), but it remained for smaller companies like 3-Com to commercialize Ethernet broadly. Hewlett-Packard commercialized the laser printer, another Xerox-PARC innovation. And so on.⁵¹

对施乐的这种批评是合理的，但并没有贬低它在 20 世纪 70 年代取得的成就。人们可以将施乐与其更敏捷的硅谷竞争对手进行比较，但出于公平起见，我们也应该将施乐与 IBM、Digital 和其他老牌计算机公司进行比较。它们中的大多数都占据了计算领域的主导地位：DEC 拥有小型计算机和交互式操作系统，Data General 拥有优雅的 Nova 架构，霍尼韦尔拥有 Multics 分时系统，Control Data 拥有 Plato 交互式系统，而 IBM 则拥有其研究实验室产生的技术创新。尽管它们没有获得所希望的回报，但这些公司都在 1980 年后为计算领域奠定了基础。

This critique of Xerox is valid but does not diminish the magnitude of what it accomplished in the 1970s. One may compare Xerox to its more nimble Silicon Valley competitors, but out of fairness one should also compare Xerox to IBM, Digital, and the other established computer companies. Most of them were in a position to dominate computing: DEC with its minicomputers and interactive operating systems, Data General with its elegant Nova architecture, Honeywell with its Multics time-sharing system, Control Data with its Plato interactive system, and IBM for the technical innovations that its research labs generated. Although they did not reap the rewards they had hoped for, each of these companies built the foundation for computing after 1980.

在 Xerox-PARC 内部，研究人员于 1973 年设计并制造了一台计算机 Alto（图 8.4）。它借鉴了麻省理工学院林肯实验室 TX-2 的架构特点，使 Alto 能够驱动复杂的屏幕和 I/O 设备，而不会严重降低处理器的性能。最终，共制造了 1000 多台，几乎所有都在公司内部使用。网络是可选的，但一旦可用，很少有 Alto 用户不使用以太网连接。制造一台 Alto 的成本约为 18,000 美元。凭借其功能，许多人声称 Alto 是第一台真正的个人计算机。然而，它并没有向公众销售——用于个人使用成本太高。52^除了使用鼠标和窗口之外，Alto 还有一个“位图”屏幕，可以通过设置 Alto 内存中的位来操作屏幕上的每个图像元素。这允许用户缩放字母以及在屏幕上混合文本和图形。这也意味着文本编辑系统将具有“所见即所得” (WYSIWYG) 的功能 - 这句话由喜剧演员 Flip Wilson 在电视节目“Laugh-In”中提出并广为人知^。53

Within Xerox-PARC, researchers designed and built a computer, the Alto, in 1973 (figure 8.4). An architectural feature borrowed from the MIT-Lincoln Labs TX-2 gave the Alto the power to drive a sophisticated screen and I/O facilities without seriously degrading the processor’s performance. Eventually over a thousand were built, and nearly all were used within the company. Networking was optional, but once available, few Alto users did without an Ethernet connection. An Alto cost about $18,000 to build. By virtue of its features, many claimed that the Alto was the first true personal computer. It was not marketed to the public, however—it would have cost too much for personal use.⁵² Besides using a mouse and windows, the Alto also had a “bit-mapped” screen, where each picture element on the screen could be manipulated by setting bits in the Alto’s memory. That allowed users to scale letters and mix text and graphics on the screen. It also meant that a text-editing system would have the feature “what you see is what you get” (WYSIWYG)—a phrase made popular by the comedian Flip Wilson on the television program “Laugh-In.”⁵³

图 8.4

施乐 Alto，约 1973 年。（来源：史密森学会。）

Figure 8.4

Xerox Alto, ca. 1973. (Source: Smithsonian Institution.)

1981 年，施乐公司推出了一款名为 8010 Star Information System 的商业版本，并在当年夏天于芝加哥举行的全国计算机大会上大张旗鼓地宣布了这一消息。广告描述了一种十年后将变得司空见惯的办公环境，甚至比 1991 年的办公室工作人员所拥有的还要强大。但这款产品失败了。大约在同一时间，施乐公司推出了一款使用 CP/M 的“普通”个人电脑，但同样未能畅销^。54

In 1981 Xerox introduced a commercial version, called the 8010 Star Information System, announced with great fanfare at the National Computer Conference in Chicago that summer. Advertisements described an office environment that would be commonplace ten years later, even more capable than what office workers in 1991 had. But the product fizzled. Around the same time Xerox introduced an “ordinary” personal computer using CP/M, but that, too, failed to sell.⁵⁴

Star 是 Alto 的衍生产品，从技术上讲，它比当时存在的几乎所有其他办公设备都要先进，包括王安 WPS。到 1984 年，个人电脑将具备 Star 的一些功能，但个人电脑集成网络要再过十年才会普及。在 20 世纪 70 年代末，王安比施乐更了解办公环境和办公需求。Star 的广告描绘了一位高管在办公桌上打电话、撰写和发送文件；施乐不知何故忘记了，企业高管甚至不会自己打电话，而是让秘书来做这件事。相比之下，王安的产品瞄准的是真正负责打字和归档的办公室工作人员。Alto 更先进，这也解释了为什么它的功能在 20 世纪 90 年代的办公计算中变得很普遍。王安更实用，但不那么先进，这也解释了王安在 20 世纪 70 年代末取得的惊人的财务成功以及随后陷入破产的原因。

The Star, derived from the Alto, was technically superior to almost any other office machine then in existence, including the Wang WPS. Personal computers would have some of the Star’s features by 1984, but integrated networks of personal computers would not become common for another ten years. In the late 1970s, Wang had a better sense than Xerox of what an office environment was like and what its needs were. Advertisements for the Star depicted an executive calling up, composing, and sending documents at his desk; somehow Xerox forgot that business executives do not even place their own telephone calls but get a secretary to do that. By contrast, Wang aimed its products at the office workers who actually did the typing and filing. The Alto was more advanced, which explains why its features became common in office computing in the 1990s. The Wang was more practical but less on the cutting edge, which explains both Wang’s stunning financial success in the late 1970s, and its slide into bankruptcy afterward.

除了发明基于 Windows 的界面外，施乐公司发明的以太网还产生了其他深远的影响。以太网提供了一种在本地环境中将计算机相互连接的有效方法。尽管个人计算的第一个十年强调将计算机用作自主、独立的设备，但到了 20 世纪 80 年代中期，通过某种基于以太网的方案在办公室中将它们连接起来已变得很普遍。最终，这种网络是一种规避格罗施定律的方法，该定律意味着一台大型昂贵的计算机的性能将优于以相同金额购买的一组小型机器。在小型计算机和个人计算机的动荡时期，这条定律一直有效；但以太网的有效性最终使它和它所支持的大型机文化衰落。55^下一章将讨论这种情况是如何发生的。

Along with its invention of a windows-based interface, Xerox’s invention of Ethernet would have other far-reaching consequences. Ethernet provided an effective way of linking computers to one another in a local environment. Although the first decade of personal computing emphasized the use of computers as autonomous, separate devices, by the mid1980s it became common to link them in offices by some form of Ethernet-based scheme. Such a network was, finally, a way of circumventing Grosch’s Law, which implied that a large and expensive computer would outperform a cluster of small machines purchased for the same amount of money. That law had held up throughout the turmoil of the minicomputer and the PC; but the effectiveness of Ethernet finally brought it, and the mainframe culture it supported, down.⁵⁵ How that happened will be discussed in the next chapter.

个人电脑：第二次浪潮，1977-1985 年

Personal Computers: the Second Wave, 1977–1985

再次，这些来自大型知名公司的自上而下的创新与来自个人电脑制造商的自下而上的创新步伐同样迅速。

Once again, these top-down innovations from large, established firms were matched by an equally brisk pace of innovation from the bottom up—from personal computer makers.

1977 年夏天，Radio Shack 开始在商店出售 TRS-80，起价为 400 美元。Model 1 使用 Z-80 芯片；它比英特尔 8080 更先进（尽管它没有复制 Altair 架构）。Model 1 包括一个键盘和一个显示器，以及用于存储的磁带。启动程序和 BASIC（不是微软的）位于只读存储器中。Radio Shack 的营销影响力及其遍布全国的商店帮助该公司一炮走红。56 由于^Radio Shack 的客户包括电子爱好者或黑客以外的人，Model 1 让个人电脑找到了大众受众。多年后，人们可以发现 TRS-80 计算机正在为小型企业做会计和盘点，例如，使用从磁带或软盘加载的简单 BASIC 程序。 TRS-80 标志着个人计算实验阶段的结束和成熟阶段的开始。

In the summer of 1977 Radio Shack began offering its TRS-80 in its stores, at prices starting at $400. The Model 1 used the Z-80 chip; it was more advanced than the Intel 8080 (although it did not copy the Altair architecture). The Model 1 included a keyboard and a monitor, and cassettes to be used for storage. A start-up routine and BASIC (not Microsoft’s) were in a read-only memory. The marketing clout of Radio Shack, with its stores all over the country, helped make it an instant hit for the company.⁵⁶ Because Radio Shack’s customers included people who were not electronics hobbyists or hackers, the Model 1 allowed the personal computer to find a mass audience. Years later one could find TRS-80 computers doing the accounting and inventory of small businesses, for example, using simple BASIC programs loaded from cassettes or a floppy disk. The TRS-80 signaled the end of the experimental phase of personal computing and the beginning of its mature phase.

同年推出的另外两台计算机完成了这一转变。Commodore PET 还配备了显示器、键盘和磁带播放器，它们都内置在一个盒子里。它使用的微处理器与英特尔 8080 的架构不同——6502（由 MOS Technologies 销售）。PET 的主要缺点是它的计算器式键盘，因此它在美国没有像当年推出的其他同类计算机那样成功。但它在欧洲卖得很好，在欧洲大陆它成为多年来的标准。

Two other computers introduced that year completed this transition. The Commodore PET also came complete with monitor, keyboard, and cassette player built into a single box. It used a microprocessor with a different architecture from the Intel 8080—the 6502 (sold by MOS Technologies). The PET’s chief drawback was its calculator-style keyboard, and for that reason it was not as successful in the U.S. as the other comparable computers introduced that year. But it sold very well in Europe, and on the Continent it became a standard for many years.

1977 年推出的第三台机器是 Apple II（图 8.5）。据说它诞生在硅谷的一个车库里，由两个理想主义的年轻人史蒂夫·乔布斯和史蒂夫·沃兹尼亚克协助完成，这已经成为硅谷的民间传说。据传说，史蒂夫·沃兹尼亚克为 Apple 选择 6502 芯片仅仅是因为它比 8080 便宜。在设计计算机之前，他曾在家酿计算机俱乐部的讨论中尝试过自己的想法，该俱乐部定期在斯坦福校园的一个大厅举行会议。Apple II 是电路设计领域的杰作。它使用的芯片比同类的 Altair 机器要少，但性能却优于大多数机器。它具有出色的彩色图形处理功能，比大多数大型机或小型计算机都要好。这使它适合于快速互动游戏，而这也是所有人都认为个人计算机擅长的少数几个方面之一。它装在一个漂亮的塑料壳里。它的名字没有威胁性，也没有什么技术性。尽管用户必须打开机箱才能连接打印机，但它并不像 Altair 系列电脑那样令人望而生畏。乔布斯、沃兹尼亚克和家酿电脑俱乐部的其他成员并没有像传说中那样发明个人电脑。但 Apple II 最接近斯图尔特·布兰德的预测，即电脑不仅会走进人们的生活，而且还会受到人们的欢迎，成为一种友好、无威胁的技术，可以丰富他们的个人生活。Apple II 的工程和设计反映了这些目标。

The third machine introduced in 1977 was the Apple II (figure 8.5). The legend of its birth in a Silicon Valley garage, assisted by two idealistic young men, Steve Jobs and Steve Wozniak, is part of the folklore of Silicon Valley. According to the legend, Steve Wozniak chose the 6502 chip for the Apple simply because it cost less than an 8080. Before designing the computer he had tried out his ideas in discussions at the Homebrew Computer Club, which met regularly at a hall on the Stanford campus. The Apple II was a tour de force of circuit design. It used fewer chips than the comparable Altair machines, yet it outperformed most of them. It had excellent color graphics capabilities, better than most mainframes or minicomputers. That made it suitable for fast-action interactive games, one of the few things that all agreed personal computers were good for. It was attractively housed in a plastic case. It had a nonthreatening, nontechnical name. Even though users had to open the case to hook up a printer, it was less intimidating than the Altair line of computers. Jobs and Wozniak, and other members of the Homebrew Computer Club, did not invent the personal computer, as the legend often goes. But the Apple II came closest to Stewart Brand’s prediction that computers would not only come to the people, they would be embraced by the people as a friendly, nonthreatening piece of technology that could enrich their personal lives. The engineering and design of the Apple II reflected those aims.

沃兹尼亚克为 Apple 编写了自己的 BASIC，但后来 Apple II 上市时推出了更好的版本，由微软为 6502 编写并以 ROM 形式提供。据说，1977 年 8 月，苹果向微软支付了 10,500 美元作为部分许可费，这笔钱在微软历史的关键时刻拯救了微软，使它免于破产。57^尽管比 TRS-80 或 PET 都贵，但 Apple II 的销量更好。没过多久，人们就为它编写了富有想象力的软件。与 Altair 一样，Apple II 也具有带扩展插槽的总线架构——沃兹尼亚克极力主张这一功能，可能是因为他在 Data General Nova 上看到了它的优势。58^总线架构使苹果和其他公司能够扩展苹果的功能，并使其在动荡的 1970 年代末到 1980 年代保持活力。 1980 年推出的卡中包括微软的 SoftCard，它允许 Apple II 运行 CP/M。对于后来以软件闻名的微软来说，这款硬件具有讽刺意味地成为当时最畅销的产品之一。

Wozniak wrote his own BASIC for the Apple, but the Apple II was later marketed with a better version, written by Microsoft for the 6502 and supplied in a ROM. A payment of $10,500 from Apple to Microsoft in August 1977, for part of the license fee, is said to have rescued Microsoft from insolvency at a critical moment of its history.⁵⁷ Although it was more expensive than either the TRS-80 or the PET, the Apple II sold better. It did not take long for people to write imaginative software for it. Like the Altair, the Apple II had a bus architecture with slots for expansion—a feature Wozniak argued strenuously for, probably because he had seen its advantages on a Data General Nova.⁵⁸ The bus architecture allowed Apple and other companies to expand the Apple’s capabilities and keep it viable throughout the volatile late 1970s and into the 1980s. Among the cards offered in 1980 was the SoftCard, from Microsoft, which allowed an Apple II to run CP/M. For Microsoft, a company later famous for software, this piece of hardware was ironically one of its best selling products at the time.

图 8.5

个人电脑：Apple II，约 1977 年，配有显示器和 Apple 磁盘驱动器。（来源：史密森学会。）

Figure 8.5

Personal computers: Apple II, ca. 1977, with a monitor and an Apple disk drive. (Source: Smithsonian Institution.)

到 1977 年底，个人计算机已经成熟。像 TRS-80 这样的机器是真正的家用电器，几乎任何人都可以购买并运行。它们对于玩游戏和学习计算基础知识很有用，但对于严肃的应用程序来说还不够好。基于 Altair 总线的系统更复杂，设置和运行起来也更困难，但如果配置得当，它可以在各种应用程序上与小型计算机竞争。Apple II 连接了这两个世界，既具有前者的灵活性，又具有后者的易用性和友好性。其基础是不断发展的商业软件行业。

By the end of 1977 the personal computer had matured. Machines like the TRS-80 were true appliances that almost anyone could buy and get running. They were useful for playing games and for learning the rudiments of computing, but they were not good enough for serious applications. Systems based on the Altair bus were more sophisticated and more difficult to set up and get running, but when properly configured could compete with minicomputers for a variety of applications. The Apple II bridged those two worlds, with the flexibility of the one and the ease of use and friendliness of the other. At the base was a growing commercial software industry.

这些都对 IBM、Digital、Data General 或 BUNCH 的计算机领域没有太大威胁。然而，几年之内，廉价的商品硬件和商业软件的强大组合将重新定义计算机行业以及依赖它的社会。直到 20 世纪 70 年代末，DEC、IBM、王安和施乐的发展轨迹与 MITS、IMSAI、苹果、Radio Shack 或其他个人电脑供应商的发展轨迹并无交集。个人电脑领域的创新似乎并不像施乐甚至 IBM 等公司的创新那么重要。但随着时间的推移，它们将对整个计算机领域产生同样大的影响。其中一项创新来自苹果。

None of this was much of a threat to the computer establishment of IBM, Digital, Data General, or the BUNCH. Within a few years, though, the potent combination of cheap commodity hardware and commercial software would redefine the computer industry and the society that would come to depend on it. The trajectories of DEC, IBM, Wang, and Xerox did not intersect those of MITS, IMSAI, Apple, Radio Shack, or the other personal computer suppliers into the late 1970s. Innovations in personal computing did not seem as significant as those at places like Xerox or even IBM. But in time they would affect all of computing just as much. One of those innovations came from Apple.

APPLE II 的磁盘驱动器和 VisiCalc

APPLE II’s Disk Drive and VisiCalc

到 1977 年，许多个人电脑公司（包括 MITS 和 IMSAI）都推出了 8 英寸软盘驱动器。这些软盘驱动器比盒式磁带好得多，但价格也昂贵。Apple II 使用盒式磁带，但到 1977 年底，史蒂夫·沃兹尼亚克 (Steve Wozniak) 正在为其设计磁盘控制器。Apple 从 Shugart Associates 购买了驱动器本身（新的 5 1/4 英寸尺寸），但沃兹尼亚克认为当时使用的控制电路过于复杂，需要多达 50 个芯片。他设计了一个使用五个芯片的电路。它曾经是、现在仍然是优雅和经济的奇迹，教授们在工程课程中将其用作范例。他后来讲述了他是如何受到美学考虑和工程考虑的驱使，使其变得简单、快速和优雅^。59

By 1977 many personal computer companies, including MITS and IMSAI, were offering 8-inch floppy disk drives. These were much better than cassette tape but also expensive. The Apple II used cassette tape, but by the end of 1977 Steve Wozniak was designing a disk controller for it. Apple purchased the drives themselves (in a new 5 1/4-inch size) from Shugart Associates, but Wozniak felt that the controlling circuits then in use were too complex, requiring as many as fifty chips. He designed a circuit that used five chips. It was, and remains, a marvel of elegance and economy, one that professors have used as an example in engineering courses. He later recounted how he was driven by aesthetic considerations as much as engineering concerns to make it simple, fast, and elegant.⁵⁹

苹果的 5 1/4 英寸软盘驱动器可以容纳 113 KB 的数据，售价 495 美元，其中包括操作系统软件和插入 Apple II 内部插槽之一的控制器。60^它非常适合个人计算机的需求 - 该驱动器允许人们营销和分发有用的商业软件，而不仅仅是简单的游戏和支票簿平衡程序，这些是磁带容量的极限。软盘存储与操作系统软件相结合，使软件生产商免受特定机器特性的影响，使软件脱颖而出。随后的几十年将继续见证硬件的进步。但由特定机器及其技术定义的计算机世代不再能够最好地描述计算的演变。除了少数例外，新计算机将不再是计算历史的关键 - 甚至不再有趣。

Apple’s 5 1/4-inch floppy drive could hold 113 Kbytes of data and sold for $495, which included operating system software and a controller that plugged into one of the Apple II’s internal slots.⁶⁰ It was a good match for the needs of the personal computer—the drive allowed people to market and distribute useful commercial software, and not just the simple games and checkbook-balancing programs that were the limit of cassette tape capacity. Floppy disk storage, combined with operating system software that insulated software producers from the peculiarities of specific machines, brought software to the fore. Ensuing decades would continue to see advances in hardware. But no longer would computer generations, defined by specific machines and their technology, best describe the evolution of computing. With a few exceptions, new computers would cease to be pivotal—or even interesting—to the history of computing.

1979 年 10 月，一款名为 VisiCalc 的程序出现在 Apple II 上。它的创建者是 Daniel Bricklin 和 Robert Frankston，两人在麻省理工学院的 MAC 项目工作时相识。Bricklin 曾在数字设备公司工作，并于 20 世纪 70 年代末进入哈佛商学院。在那里，他接触到了一代又一代商学院学生必须掌握的计算：对行和列数据进行算术运算，通常计算公司在几个月、几个季度或几年内的业绩。这种计算在整个金融界很常见，几十年来一直使用 IBM 打孔卡设备半自动化完成。他回忆起他的一位教授在黑板上张贴、更改和分析这些表格，使用他的助手前一天晚上手工计算的数字。Bricklin 构思了一个程序来自动化这些“电子表格”（会计师中已经很少使用这个术语）。拥有自己的小型软件营销公司的二年级学生 Dan Flystra 同意帮助他营销这个程序。布里克林随后前往弗兰克斯顿，后者同意帮助撰写。

In October 1979 a program called VisiCalc was offered for the Apple II. Its creators were Daniel Bricklin and Robert Frankston, who had met while working on Project MAC at MIT. Bricklin had worked for Digital Equipment Corporation and in the late 1970s attended the Harvard Business School. There he came across the calculations that generations of B-school students had to master: performing arithmetic on rows and columns of data, typically of a company’s performance for a set of months, quarters, or years. Such calculations were common throughout the financial world, and had been semi-automated for decades using IBM punched-card equipment. He recalled one of his professors posting, changing, and analyzing such tables on the blackboard, using figures that his assistant had calculated by hand the night before. Bricklin conceived of a program to automate these “spreadsheets” (a term already in limited use among accountants). Dan Flystra, a second-year student who had his own small software marketing company, agreed to help him market the program. Bricklin then went to Frankston, who agreed to help write it.

1979 年 1 月，布里克林和弗兰克斯顿在马萨诸塞州阿灵顿弗兰克斯顿的阁楼上成立了 Software Arts 公司（波士顿地区的车库比硅谷少）。那年春天，弗兰克斯顿和布里克林租用了 MIT Multics 系统，这个项目开始成型。6 月，VisiCalc 在全国计算机大会上亮相。这个名字代表了可视计算器，尽管它的灵感可能来自于一天早上在马萨诸塞大道上的 Vic's Egg on One 咖啡店吃早餐。（内森·普里蒂金不会同意，但这样的餐馆是波士顿场景的另一个常见特征，在硅谷却找不到。）⁶¹

In January 1979 Bricklin and Frankston formed Software Arts, based in Frankston’s attic in Arlington, Massachusetts (the Boston area has fewer garages than in Silicon Valley). That spring the program took shape, as Frankston and Bricklin rented time on the MIT Multics system. In June, VisiCalc was shown at the National Computer Conference. The name stood for visible calculator, although inspiration for it may have come from eating breakfast one morning at Vic’s Egg on One coffee shop on Massachusetts Avenue. (Nathan Pritikin would not have approved, but such eateries are another common feature of the Boston scene not found in Silicon Valley.)⁶¹

布里克林想为 DEC 的设备开发这个程序，“也许可以在 128 号公路上挨家挨户推销”。弗莱斯特拉有一台 Apple II 和一台 TRS-80；他让布里克林使用 Apple，因此 VisiCalc 是在 Apple 上开发的。价格约为 200 美元。Apple 本身对营销该程序不感兴趣。但该产品获得了好评。一位金融分析师说，这可能是“软件尾巴摇硬件狗”。62^他说得对：在许多计算机商店里，人们会进来询问 VisiCalc，然后询问运行它所需的计算机（Apple II）。到 1981 年中期（IBM 个人计算机发布的那一年，这一事件导致了 Software Arts 的倒闭），销量突破了十万大关。

Bricklin wanted to develop this program for DEC equipment, “and maybe sell it door-to-door on Route 128.” Flystra had an Apple II and a TRS-80; he let Bricklin use the Apple, so VisiCalc was developed on an Apple. The price was around $200. Apple itself was not interested in marketing the program. But the product received good reviews. A financial analyst said it might be the “software tail that wags the hardware dog.”⁶² He was right: in many computer stores people would come in and ask for VisiCalc and then the computer (Apple II) they needed to run it. Sales passed the hundred thousand mark by mid-1981 (the year the IBM personal computer was announced, an event that led to Software Arts’s demise).

Apple II 的拥有者现在可以做两件事，即使是那些拥有大型机的人也无法做到。第一是玩游戏；诚然，这不是一个严肃的应用程序，但仍然拥有健康的市场。第二是使用 VisiCalc；它与在大型机上运行的任何应用程序一样重要。文字处理，以前只有能够买得起 Wang 或 Lanier 系统的企业客户才能使用，很快就跟上了。

An owner of an Apple II could now do two things that even those with access to mainframes could not do. The first was play games; admittedly not a serious application, but one that nevertheless had a healthy market. The second was use VisiCalc; which was as important as any application running on a mainframe. Word processing, previously available only to corporate customers who could afford systems from Wang or Lanier, soon followed.

IBM PC（1981年）

IBM PC (1981)

尽管在 Apple II 及其软盘驱动器面世后，人们可以说硬件进步不再推动计算机历史的发展，但也有一些例外，其中之一就是 IBM 个人计算机。它在 1981 年 8 月的发布确实很重要，尽管它代表了对现有技术的渐进式进步。它的处理器是英特尔 8088，源自 8080，内部以 16 位字处理数据（外部通信仍为 8 位）。63^它使用 ASCII 代码。它的 62 针总线架构类似于 Altair 的总线，并带有五个空的扩展槽。Microsoft BASIC 以只读存储器芯片的形式提供。它有一个内置的卡带端口，与 BASIC 相结合，意味着不需要磁盘操作系统。大多数客户想要磁盘存储，他们有三种操作系统可供选择：CP/M-86（加州大学圣地亚哥分校设计的基于 Pascal 的系统）和 Microsoft 的 PC-DOS。 CP/M-86 直到 1982 年才准备就绪，而且很少有客户购买 Pascal 系统，因此 PC-DOS 占了上风。软盘驱动器、键盘和视频监视器也是以前使用的组件的变体。IBM 将监视器驱动程序集成到 PC 的基本电路板中，这样用户就不必占用通信端口。单色监视器可以显示 25 行 80 个字符的全屏 - 这是对 Apple II 的改进，对于严肃的办公应用程序来说是必不可少的。还有带彩色监视器的版本（图 8.6）。

Although after the Apple II and its floppy drive were available, one could say that hardware advances no longer drove the history of computing, there were a few exceptions, and among them was the IBM Personal Computer. Its announcement in August 1981 did matter, even though it represented an incremental advance over existing technology. Its processor, an Intel 8088, was descended from the 8080, handling data internally in 16-bit words (external communication was still 8 bits).⁶³ It used the ASCII code. Its 62-pin bus architecture was similar to the Altair’s bus, and it came with five empty expansion slots. Microsoft BASIC was supplied in a read-only memory chip. It had a built-in cassette port, which, combined with BASIC, meant there was no need for a disk operating system. Most customers wanted disk storage, and they had a choice of three operating systems: CP/M-86, a Pascal-based system designed at the University of California at San Diego, and PC-DOS from Microsoft. CP/M-86 was not ready until 1982, and few customers bought the Pascal system, so PC-DOS prevailed. The floppy disk drives, keyboard, and video monitor were also variants of components used before. IBM incorporated the monitor driver into the PC’s basic circuit board, so that users did not tie up a communication port. The monochrome monitor could display a full screen of 25 lines of 80 characters— an improvement over the Apple II and essential for serious office applications. A version with a color monitor was also available (figure 8.6).

随着 PC 的推出，IBM 还宣布推出文字处理、会计、游戏软件和 VisiCalc 版本。1982 年 10 月推出的电子表格 1-2-3 由 Lotus Development 开发，它利用了 PC 的架构，运行速度比其竞争对手 VisiCalc 快得多。IBM 个人电脑和 Lotus 1-2-3 的组合很快在销量上超过了苹果，打消了人们对这些机器作为大型机和小型机的强劲竞争对手的疑虑。1982 年 12 月，《时代》杂志将这款电脑评为 1983 年“年度机器” ^。64

With the PC, IBM also announced the availability of word processing, accounting, games software, and a version of VisiCalc. A spreadsheet introduced in October 1982, 1-2-3 from Lotus Development, took advantage of the PC’s architecture and ran much faster than its competitor, VisiCalc. This combination of the IBM Personal Computer and Lotus 1-2-3 soon overtook Apple in sales and dispelled whatever doubts remained about these machines as serious rivals to mainframe and minicomputers. In December 1982 Time magazine named the computer “Machine of the Year” for 1983.⁶⁴

图 8.6

个人计算机：IBM PC，1981 年。请注意两个内置软盘驱动器。（来源：史密森学会。）

Figure 8.6

Personal computers: IBM PC, 1981. Note the two internal floppy disk drives. (Source: Smithsonian Institution.)

操作系统

MS-DOS

当 IBM 位于佛罗里达州博卡拉顿的分部着手开展代号为“Chess”的项目时，微软还是一家小公司。微软以其 BASIC 版本而闻名。IBM 为一款名为 System/23 Datamaster 的产品开发了一个 BASIC 版本，但由于需要将此版本的 BASIC 与其他 IBM 产品进行协调，导致开发延迟。Chess 团队看到了个人电脑领域正在发生的事情，他们意识到任何延迟都将是致命的。因此，他们将几乎该产品的每个部分（包括软件）都外包给 IBM 组织^。65

Microsoft was a small company when an IBM division in Boca Raton, Florida, embarked on this project, code named “Chess.” Microsoft was best known for its version of BASIC. IBM had developed a version of BASIC for a product called the System/23 Datamaster, but the need to reconcile this version of BASIC with other IBM products caused delays. The Chess team saw what was happening in the personal computer field, and they recognized that any delays would be fatal. As a result they would go outside the IBM organization for nearly every part of this product, including the software.⁶⁵

1980 年夏天，IBM 代表与比尔·盖茨接洽，希望他能提供可以在 IBM 选择的 Intel 8088 上运行的 BASIC 版本。66 ^IBM认为它能够使用 CP/M 的一个版本作为操作系统；CP/M 已经确立为基于 8080 的系统的标准，而 Digital Research 正在开发 16 位扩展。但与 Digital Research 的 Gary Kildall 的谈判陷入僵局。当 IBM 访问 Digital Research 达成交易时，Kildall 不在场，而负责公司行政工作的他的妻子拒绝签署 IBM 的保密协议。（考虑到多年来对 IBM 的指控，她的说法并非不合理。67 ^）无论如何，Digital Research 的 16 位 CP/M 版本尚未开发到位，尽管该公司已经承诺了一段时间。（在 PC-DOS 占据主导地位后，它最终被提供给 IBM PC。）

Representatives of IBM approached Bill Gates in the summer of 1980 to supply a version of BASIC that would run on the Intel 8088 that IBM had chosen.⁶⁶ IBM thought it would be able to use a version of CP/M for the operating system; CP/M was already established as the standard for 8080-based systems, and Digital Research was working on a 16-bit extension. But negotiations with Gary Kildall of Digital Research stalled. When IBM visited Digital Research to strike the deal, Kildall was not there, and his wife, who handled the company’s administrative work, refused to sign IBM’s nondisclosure agreement. (Given the charges that had been leveled against IBM over the years, she was not being unreasonable.⁶⁷) In any event, Digital Research’s 16-bit version of CP/M was not far enough along in development, although the company had been promising it for some time. (It was eventually offered for the IBM PC, after PC-DOS had become dominant.)

最后，微软向 IBM 提供了自己的 16 位操作系统。IBM 将其称为 PC-DOS，而微软可以自由地将其作为 MS-DOS 推广到其他地方。PC-DOS 基于 86-DOS，这是西雅图计算机产品公司的 Tim Paterson 为 8086 芯片编写的操作系统。微软最初支付了约 15,000 美元，以购买使用西雅图计算机产品公司作品的权利。（微软后来支付了更大的金额以购买全部权利。）西雅图计算机产品公司在内部将其称为代号 QDOS，即“快速而粗略的操作系统”；它最终被称为 MS-DOS，这是有史以来寿命最长、最具影响力的软件之一^。68

In the end, Microsoft offered IBM a 16-bit operating system of its own. IBM called it PC-DOS, and Microsoft was free to market it elsewhere as MS-DOS. PC-DOS was based on 86-DOS, an operating system that Tim Paterson of Seattle Computer Products had written for the 8086 chip. Microsoft initially paid about $15,000 for the rights to use Seattle Computer Products’s work. (Microsoft later paid a larger sum of money for the complete rights.) Seattle Computer Products referred to it internally by the code name QDOS for “Quick and Dirty Operating System”; it ended up as MS-DOS, one of the longest-lived and most-influential pieces of software ever written.⁶⁸

MS-DOS 秉承了 CP/M 的精神。与传说相反，它不仅仅是为先进的 8086 芯片编写的 CP/M 扩展。帕特森熟悉 Cromemco 个人计算机使用的 CP/M 方言，以及 Northstar 和 Altair 的其他几个后代提供的操作系统。CP/M 用户手册是另一个影响因素，尽管帕特森无法访问 CP/M 源代码。另一个影响因素是 Microsoft BASIC 的高级版本，它也支持磁盘存储，这可能导致 MS-DOS 使用文件分配表来跟踪磁盘上的数据。86-DOS 确实使用了与 CP/M 相同的内部函数调用；实际上，它使用了英特尔在记录芯片时发布的 8086 地址和约定，以便于在新的微处理器上运行为 8080 编写的程序。它使用了 CP/M 命令“Type”、“Rename”和“Erase”。 MS-DOS 还保留了 CP/M 的 BIOS 概念，这使得它只需进行相对较小的改动即可在不同制造商的计算机上运行^。69

MS-DOS was in the spirit of CP/M. Contrary to folklore, it was not simply an extension of CP/M written for the advanced 8086 chip. Paterson was familiar with a dialect of CP/M used by the Cromemco personal computer, as well as operating systems offered by Northstar and a few other descendants of the Altair. A CP/M users manual was another influence, although Paterson did not have access to CP/M source code. Another influence was an advanced version of Microsoft BASIC that also supported disk storage, which probably led to the use of a file allocation table by MS-DOS to keep track of data on a disk. The 86-DOS did use the same internal function calls as CP/M; actually, it used 8086 addresses and conventions that Intel had published in documenting the chip, to make it easy to run programs written for the 8080 on the new microprocessor. It used the CP/M commands “Type,” “Rename,” and “Erase.” MS-DOS also retained CP/M’s notion of the BIOS, which allowed it to run on computers from different manufacturers with relatively minor changes.⁶⁹

值得一提的是 CP/M 和 MS-DOS 之间的差异，因为这些差异有助于解释后者的成功。一些变化相对较小：神秘的万能 PIP 命令被改为更平淡无奇的术语，如“复制”；这使 MS-DOS 更容易被新一代计算机用户接受，但切断了与数字设备公司的历史联系，而数字设备公司的软件是个人计算机系统的真正祖先。CP/M 的语法将第一个参数指定为目标，将第二个参数指定为源；这被反转为对大多数人来说似乎更自然的东西。（英特尔的汇编代码和 IBM System/360 的汇编程序也使用了 CP/M 语法）。70^更根本的改进包括 MSDOS 能够寻址更多内存 - 这是它为英特尔芯片编写的结果。MS-DOS 使用文件分配表；CP/M 使用了一种不太复杂的方法。如果将错误的磁盘插入驱动器，CP/M 需要重新启动系统这一烦人的事情被消除了。在 MSDOS 中执行此操作会产生一条消息，“中止，重试，失败？”这条消息后来被引用为 MS-DOS 不友好用户界面的一个例子，但说这话的人可能从未经历过 CP/M 的“热启动”消息，这更糟糕，有时给人一种被真正的启动踢的感觉。一些功能可能受到 UNIX 的启发，例如版本 2，它允许用户将文件存储在磁盘上的目录和子目录的分层树中。71 ^Tim Paterson 后来表示，他原本打算将多任务功能纳入 DOS，但“他们 [Microsoft] 需要快速获得一些东西。” ⁷²

It is worth mentioning the differences between CP/M and MS-DOS, since these help explain the latter’s success. A few changes were relatively minor: the cryptic all-purpose PIP command was changed to more prosaic terms like “Copy”; this made MS-DOS more accessible to a new generation of computer users but severed the historical link with the Digital Equipment Corporation, whose software was the real ancestor of personal computer systems. CP/M’s syntax specified the first argument as the destination and the second as the source; this was reversed to something that seems to be more natural to most people. (The CP/M syntax was also used by Intel’s assembler code and by the assembler for the IBM System/360).⁷⁰ More fundamental improvements included MSDOS’s ability to address more memory—a consequence of the Intel chip it was written for. MS-DOS used a file allocation table; CP/M used a less-sophisticated method. CP/M’s annoying need to reboot the system if the wrong disk was inserted into a drive was eliminated. Doing that in MSDOS produced a message, “Abort, Retry, Fail?” This message would later be cited as an example of MS-DOS’s unfriendly user interface, but those who said that probably never experienced CP/M’s “Warm Boot” message, which was much worse and sometimes gave the feeling of being kicked by a real boot. Several features may have been inspired by UNIX, for example, version 2, which allowed users to store files on a disk in a hierarchical tree of directories and subdirectories.⁷¹ Tim Paterson later stated that he had intended to incorporate multitasking into DOS, but “they [Microsoft] needed to get something really quick.”⁷²

系统软件，无论是用于大型机还是个人计算机，似乎总是需要“数不清的人工”来创建，超出预算，并且充斥着大量低效的代码。Tim Paterson 最初在 86-DOS 上的工作花了大约两个月的时间，代码占用了大约 6 K。73 ^MS -DOS 曾经是，现在仍然是一个熟练的程序。它是麻省理工学院 TX-0 开始的交互式计算思想的顶峰。它有缺点，有些可能很严重，但那些声称 MS-DOS 的成功完全归功于比尔盖茨的狡猾，或者当 IBM 代表来找他时 Gary Kildall 驾驶他的飞机的人是错误的。

System software, whether for mainframes or for personal computers, seems always to require “mythical man-months” to create, to come in over budget, and to be saddled with long passages of inefficient code. Tim Paterson’s initial work on 86-DOS took about two months, and the code occupied about 6 K.⁷³ MS-DOS was, and is, a piece of skillful programming. It was the culmination of ideas about interactive computing that began with the TX-0 at MIT. It has its faults, some perhaps serious, but those who claim that MS-DOS’s success was solely due to Bill Gates’s cunning, or to Gary Kildall’s flying his airplane when IBM’s representatives came looking for him, are wrong.

PC 和 IBM

The PC and IBM

个人电脑是 IBM 继 5100 之后第二次进军该市场，甚至在一些产品文献中将其命名为 5150。无论是 IBM 还是其他任何人都没有预见到它会取得如此大的成功，也没有预见到其他人会复制其架构，使其成为未来十年及以后的标准。为了保持计算机行业的悠久传统，IBM 严重低估了销量：它估计总共只有 250,000 台；“事实证明，有几个月^我们制造和销售了近这么多的系统。74 MS-DOS 使微软从一家主要销售 BASIC 的公司转变为一家在操作系统领域主导小型系统行业的公司。IBM 发现自己拥有一款非常成功的产品，该产品由其他人设计的部件组成，使用 ASCII 而不是 EBCDIC，并且操作系统没有完全使用权。据说，如果 IBM 的个人电脑部门是一家独立的公司，那么它在 1984 年的行业排名将排在 IBM 和数字设备公司之后，位列第三。十年内，安装了超过五千万台计算机，这些计算机是原始 PC 架构的变体，并运行 MS-DOS 的高级版本^。75

The Personal Computer was IBM’s second foray into this market, after the 5100—it even had the designation 5150 in some product literature. Neither IBM nor anyone else foresaw how successful it would be, or that others would copy its architecture to make it the standard for the next decade and beyond. In keeping with a long tradition in the computer industry, IBM grossly underestimated sales: it estimated a total of 250,000 units; “[a]s it turned out, there were some months when we built and sold nearly that many systems.⁷⁴ MS-DOS transformed Microsoft from a company that mainly sold BASIC to one that dominated the small systems industry in operating systems. IBM found itself with an enormously successful product made up of parts designed by others, using ASCII instead of EBCDIC, and with an operating system it did not have complete rights to. It was said that if IBM’s Personal Computer division were a separate company, it would have been ranked #3 in the industry in 1984, after the rest of IBM and Digital Equipment Corporation. Within ten years there were over fifty million computers installed that were variants of the original PC architecture and ran advanced versions of MS-DOS.⁷⁵

“更好是好的敌人”

“The Better is the Enemy of the Good”

技术产品的进化经常被拿来与自然选择的生物进化相比较。两者有很多相似之处，包括市场选择力量如何影响技术的生存。76 两者^也有区别：生物从其父母那里继承了特性（最多两个），但发明家可以从任意数量的现有设备中借用东西。大自然也没有西摩·克雷那样的特权，即在开始设计一款新的电脑时从一张白纸开始。

The evolution of technological artifacts is often compared to the evolution by natural selection of living things. There are many parallels, including the way selective forces of the marketplace affect the survival of a technology.⁷⁶ There are differences, too: living things inherit their characteristics from their parents—at most two—but an inventor can borrow things from any number of existing devices. Nor does nature have the privilege that Seymour Cray had, namely, to start with a clean sheet of paper when embarking on a new computer design.

个人计算的历史表明，这些差异可能比想象的要小。IBM PC 的微处理器源自为终端设计的芯片，尽管 Datapoint 从未将其用于终端。它的操作系统源自一个“快速而粗糙”的操作系统，最初只是权宜之计。PC 的直接可寻址内存限制为 640 K。这也是计划外的，与英特尔微处理器的固有限制无关。640 K 被认为远远绰绰有余；几年之内，这个限制就成了程序员和用户的负担。IBM PC 及其克隆产品让商业软件脱颖而出，只要它能在该计算机或 100% 兼容的机器上运行。那些预测并渴望这一刻的梦想家现在心情复杂。这是他们想要的，但他们没有预料到要付出的代价，即被困在 IBM PC 及其操作系统的架构中。

The history of personal computing shows that these differences are perhaps less than imagined. The IBM PC’s microprocessor descended from a chip designed for a terminal, although Datapoint never used it for that. Its operating system descended from a “quick and dirty” operating system that began as a temporary expedient. The PC had a limit of 640 K of directly addressable memory. That, too, was unplanned and had nothing to do with the inherent limits of the Intel microprocessor. 640 K was thought to be far more than adequate; within a few years that limit became a millstone around the necks of programmers and users alike. The IBM PC and its clones allowed commercial software to come to the fore, as long as it could run on that computer or machines that were 100 percent compatible with it. Those visionaries who had predicted and longed for this moment now had mixed feelings. This was what they wanted, but they had not anticipated the price to be paid, namely, being trapped in the architecture of the IBM PC and its operating system.

Macintosh（1984年）

Macintosh (1984)

在那些看着 IBM PC 并问为什么不做更好的事情的人中，有一群苹果员工。他们嘲笑其保守的设计，忘记了 IBM 是经过深思熟虑后决定生产一款革命性的机器。他们看到了 MS-DOS 的局限性，但没有看到它作为标准的价值。（当然，当时 IBM 也没有看到。）但是，如果个人计算机采用了施乐公司帕洛阿尔托研究中心在过去十年进行的一些研究，它会是什么样子？施乐之星在 PC 问世几个月后就发布了，但未能流行起来。苹果的一些人认为他们可以更成功。

Among those who looked at the IBM PC and asked why not something better were a group of people at Apple. They scoffed at its conservative design, forgetting that IBM had made a deliberate decision to produce an evolutionary machine. They saw the limitations of MS-DOS, but not its value as a standard. (Of course, neither did IBM at the time.) But what would personal computing be like if it incorporated some of the research done in the previous decade at Xerox’s Palo Alto Research Center? The Xerox Star had been announced within months of the PC, but it failed to catch on. Some people at Apple thought they could be more successful.

尽管 20 世纪 70 年代 Xerox-PARC 开展了众多创造性活动，但必须强调的是，个人计算的根源——微处理器、Altair、总线架构、Apple II、BASIC、CP/M、VisiCalc、IBM PC、软盘、Lotus 1-2-3 和 MS-DOS——都与 Xerox-PARC 的研究毫无关系。

For all the creative activity that went on at Xerox-PARC in the 1970s, it must be emphasized that the roots of personal computing—the microprocessor, the Altair, the bus architecture, the Apple II, BASIC, CP/M, VisiCalc, the IBM PC, the floppy disk, Lotus 1-2-3, and MS-DOS—owed nothing to Xerox-PARC research.

1979 年，情况开始发生变化。那年秋天，苹果开始研发一款名为 Macintosh 的电脑。这款电脑是杰夫·拉斯金 (Jef Raskin) 的创意，他在加入苹果之前曾是加州大学圣地亚哥分校的计算机科学教授。他还曾是一家小型计算机中心的负责人，在那里教学生编写 Data General Novas 程序。77^拉斯金还曾是斯坦福大学人工智能实验室的访问学者，在那里他熟悉了施乐-PARC 的情况。据拉斯金说，他说服当时正在开发另一款基于文本的计算机的苹果团队将他在 PARC 看到的图形功能融入其中。苹果于 1983 年推出了这款电脑 Lisa。与施乐之星一样，它价格昂贵（约 10,000 美元），销量令人失望。拉斯金的 Macintosh 将保留 Lisa 的最佳功能，但售价是 Apple II 客户可以承受的。⁷⁸就像计算机历史上的许多事情一样，关于谁应该为 Macintosh 负责也存在争议。79许多历史记载，苹果联合创始人史蒂夫·乔布斯¹⁹⁷⁹年访问 PARC 是将 PARC 技术转移到大众市场的关键时刻。在那次访问之前，苹果已经在着手 Macintosh 的研发。不过，这次访问确实导致乔布斯从施乐挖走了几名关键人物，而人员流动是技术转移的最佳方式。据拉斯金称，这次访问还导致乔布斯坚持要求 Macintosh 具有原始设计中没有的功能。其中之一就是鼠标（图 8.7）^。80

In 1979 that began to change. That fall Apple began work on a computer called the Macintosh. It was the brainchild of Jef Raskin, who before joining Apple had been a professor of computer science at UC San Diego. He had also been the head of a small computer center, where he taught students to program Data General Novas.⁷⁷ Raskin had also been a visiting scholar at Stanford’s Artificial Intelligence Laboratory, and while there he became familiar with what was going on at Xerox-PARC. According to Raskin, he persuaded the Apple team then developing another text-based computer to incorporate the graphics features he had seen at PARC. Apple introduced that computer, the Lisa, in 1983. Like the Xerox Star, it was expensive (around $10,000), and sales were disappointing. Raskin’s Macintosh would preserve the Lisa’s best features but sell at a price that Apple II customers could afford.⁷⁸ As with so much in the history of computing, there is a dispute over who was responsible for the Macintosh.⁷⁹ Many histories describe a visit by Apple cofounder Steve Jobs to PARC in 1979 as the pivotal moment in transferring PARC technology to a mass market. Work on the Macintosh was already underway at Apple by the time of that visit. The visit did result in Jobs’ hiring several key people away from Xerox, however, and moving people is the best way to transfer technology. According to Raskin, the visit also resulted in Jobs’ insisting that the Macintosh have features not present in the original design. Among those was the mouse (figure 8.7).⁸⁰

图 8.7

个人电脑：Apple Macintosh，1984 年。大多数 Macintosh 用户很快发现，该机器需要第二个外部磁盘驱动器。（来源：史密森学会。）

Figure 8.7

Personal computers: Apple Macintosh, 1984. Most Macintosh users soon found that the machine required a second, external disk drive. (Source: Smithsonian Institution.)

1984 年 1 月，苹果在超级碗期间推出了 Macintosh，并发布了一则传奇广告。广告中，苹果承诺 Macintosh 将阻止 1984 年成为奥威尔小说《1984》所预言的技术反乌托邦。这款电脑的售价为 2,495 美元，高于拉斯金所定下的 1,000 美元，但比 Lisa 便宜。它比 IBM PC 贵，但当时没有一台 PC（无论用户添加什么软件或主板）可以提供 Macintosh 的图形界面。

In January 1984 Apple introduced the Macintosh in a legendary commercial during the Super Bowl, in which Apple promised that the Macintosh would prevent the year 1984 from being the technological dystopia forecast by Orwell’s novel 1984. The computer sold for $2,495—more than the $1,000 Raskin was aiming for, but cheaper than the Lisa. It was more expensive than an IBM PC, but no PC at that time, no matter what software or boards users added, could offer the graphical interface of the Macintosh.

Macintosh 使用 Motorola 68000 微处理器，其架构与 PDP-11 类似。该计算机配备单个磁盘驱动器，采用新的 3 1/2 英寸格式，高分辨率黑白显示器，鼠标和 128K 内存。大多数用户发现他们很快就必须升级到 512K“Fat Mac”；他们还发现有必要购买第二个磁盘驱动器。同时发布了几个程序：一个“绘画”（绘图）程序，基于 Xerox-PARC 在 Data General Nova 上所做的工作，以及一个接近所见即所得的文字处理器。

The Macintosh used a Motorola 68000 microprocessor, whose architecture resembled that of the PDP-11. The computer came with a single disk drive, using the new 3 1/2-inch form, a high-resolution black-onwhite monitor, a mouse, and 128K of memory. Most users found they soon had to upgrade to a 512K “Fat Mac”; they also found it necessary to purchase a second disk drive. A few programs were announced at the same time: a “paint” (drawing) program, based on work done at Xerox-PARC on a Data General Nova, and a word processor that came close to WYSIWYG.

一年后，Macintosh 配备了基本的网络功能，称为 AppleTalk。这允许简单地共享文件和打印机。与该系统的许多功能一样，它简单易用，多年来一直没有受到 PC 及其克隆产品的挑战。但没有硬盘选项，因此用户无法有效地将 Mac 设置为其他产品的服务器。在家中使用 Macintosh 的人不会连接到网络，而且 Mac 无法在办公环境中挑战 IBM 及其克隆产品的领先地位，除非在那些特别需要图形功能的办公室。与 Apple II 和 IBM PC 不同，Macintosh 是“封闭的”：用户无法添加主板，甚至不鼓励打开机箱。81^这是一个大胆的（有人认为是愚蠢的）偏离主流智慧的做法，但它有助于使 Macintosh 比 Lisa 或 Star 更便宜、更小、更快。 1987 年推出的版本提供了彩色功能并开放了系统，尽管苹果仍然严格控制着 Macintosh 的配置^。82

A year later the Macintosh came with a rudimentary networking ability, called AppleTalk. This allowed the simple sharing of files and printers. Like so much about the system, it was simple, easy to use, and not challenged by the PC and its clones for years. But there was no hard disk option, so users could not effectively set up a Mac as a server to the others. A person using a Macintosh at home would not be connected to a network, and the Mac was unable to challenge the lead of IBM and its clones in an office environment, except in those offices where the graphics abilities were especially needed. Unlike the Apple II and the IBM PC, the Macintosh was “closed”: users could not add boards and were discouraged from even opening up the case.⁸¹ This was a bold— some argued foolish—departure from the prevailing wisdom, but it helped make the Macintosh cheaper, smaller, and faster than the Lisa or the Star. A version introduced in 1987 offered color and opened up the system, although Apple still tightly controlled the Macintosh’s configuration.⁸²

Mac 优雅的系统软件是其最伟大的成就。它展示了极为罕见的美学美感和实用工程的结合。人们可以指出具体的细节。当打开或关闭文件时，其符号在屏幕上以小步扩展或收缩 - 不知何故感觉恰到好处。归根结底，这种感觉是主观的，但很少有人会不同意。Macintosh 软件在工程产品中是罕见的。该系统随着 Mac 的发展而发展，并受到微软的最高赞扬，微软试图用其 Windows 程序复制它。人们可以希望未来的某个系统也能拥有这种组合，但可能性并不大。

The Mac’s elegant system software was its greatest accomplishment. It displayed a combination of aesthetic beauty and practical engineering that is extremely rare. One can point to specific details. When a file was opened or closed, its symbol expanded or contracted on the screen in little steps—somehow it just felt right. Ultimately this feeling is subjective, but it was one that few would disagree with. The Macintosh software was something rarely found among engineering artifacts. The system evolved as the Mac grew, and it was paid the highest compliment from Microsoft, who tried to copy it with its Windows program. One can hope that some future system will have that combination as well, but the odds are not in favor of it.

Macintosh 比 Alto 功能更强大，运行速度比 Lisa 更快，但其软件占用的内存仅为这两个前辈的一小部分。它不仅仅是 Xerox 在 PARC 所做工作的复制品。但如此创新是有代价的：程序员很难为 Macintosh 开发应用程序软件，尤其是与 MS-DOS 相比。虽然比 Lisa 快，但其复杂的图形意味着它无法像 DOS 程序（如 Lotus 1-2-3）那样快，后者使用更接近机器码的原始命令。在经验丰富的客户中，这产生了分歧：一组人喜欢 Mac 的优雅和精致，而其他人则更喜欢 MSDOS 允许的原始马力和对单个位的访问。对于那些不是计算机神职人员的人来说，Macintosh 是天赐之物；由于其相对较慢而浪费的任何时间都被用户不必花时间阅读难以理解的用户手册所弥补。

The Macintosh had more capability than the Alto, it ran faster than the Lisa, yet its software occupied a fraction of the memory of either of those predecessors. It was not just a copy of what Xerox had done at PARC. But there was a price for being so innovative: the Macintosh was difficult for programmers to develop applications software for, especially compared to MS-DOS. And though faster than the Lisa, its complex graphics meant that it could not be as fast as a DOS program, like Lotus 1-2-3, that used more primitive commands that were closer to machine code. Among sophisticated customers that created a split: one group favored the elegance and sophistication of the Mac, while others preferred the raw horsepower and access to individual bits that MSDOS allowed. For those who were not members of the computer priesthood, the Macintosh was a godsend; whatever time was lost by its relative slowness was more than compensated for by the time the user did not have to spend reading an indecipherable users manual.

微软曾为 Macintosh 提供过一些应用软件，但苹果公司则自行开发并控制其操作系统。在 Macintosh 发布之前，其他公司就已尝试为 IBM PC 提供类似的界面。1982 年，VisiCalc 的创建者宣布为 IBM PC 推出了一款名为 VisiOn 的产品，该产品与 Macintosh 的界面类似，但从未兑现承诺。IBM 开发了一款名为 Top View 的程序，Digital Research 则开发了类似的 GEM（图形环境管理器）。微软推出了一款名为 Interface Manager 的产品，但 20 世纪 80 年代中期推出的早期版本销量不佳。Interface Manager 的后续版本更名为“Windows”，取得了巨大成功。Windows 的第 3 版（突破性版本）直到 1990 年左右才推出，因此在接下来的七年里，IBM PC 及其克隆产品都以从小型计算机世界继承下来的原始 MS-DOS 界面而闻名。

Microsoft had supplied some of the applications software for the Macintosh, but Apple developed and controlled its operating system in-house. Even before the Macintosh’s announcement, other companies were trying to provide a similar interface for the IBM PC. In 1982 the creators of VisiCalc announced a product called VisiOn for the IBM PC that was similar to the Macintosh’s interface but never lived up to its promise. IBM developed a program called Top View, and Digital Research developed GEM (Graphics Environment Manager) along the same lines. Microsoft came up with a product called Interface Manager, but early versions introduced in the mid-1980s sold poorly. Later versions of Interface Manager, renamed “Windows,” would succeed dramatically. Version 3 of Windows, the breakthrough version, was not introduced until around 1990, so for the next seven years, IBM PCs and their clones would be known by the primitive MS-DOS interface inherited from the minicomputer world.

与 IBM PC 一样，Macintosh 的设计也为内存扩展设置了障碍，只不过它有 4 兆字节，而不是 PC 的 640 千字节。1985 年推出的激光打印机完成了 Xerox-PARC 创新的转移，使 Macintosh 至少在某些办公室中占据了稳固的地位。Macintosh 的 VisiCalc 对应程序是 Aldus 于 1985 年推出的一款名为 PageMaker 的程序。当与激光打印机结合使用时，它允许用户在 Apple 上进行复杂的打印，而成本仅为传统方法的一小部分。

Like the IBM PC, the Macintosh’s design created a barrier to expanding memory, only it was a more generous 4 megabytes instead of the PC’s miserly 640 Kbytes. A laser printer offered in 1985 completed the transfer of Xerox-PARC innovations and allowed the Macintosh to keep a strong foothold in at least some offices. The Macintosh’s equivalent of VisiCalc was a program called PageMaker from Aldus, introduced in 1985. When combined with the laser printer it allowed users to do sophisticated printing on an Apple, at a fraction of the cost of traditional methods.

克隆人

The Clones

个人计算机革命似乎与之前的大型机时代没有多大关系，但随着时间的推移，我们可以发现一些共同的主题。IBM 凭借其 System/360 取得了成功，并且需要提供大量有关它的技术信息，这导致了即插即用兼容行业的出现，而这又导致 IBM 不得不调整自己的产品线。个人计算机也发生了类似的事情，只是这次的结果不同。大多数 IBM 个人计算机（包括 8088 微处理器）都由其他制造商生产的部件组成，这些制造商可以自由地将这些部件出售给其他地方。例如，微软保留了向其他制造商出售其操作系统的权利。个人计算机之所以成为“IBM PC”，其核心是基本输入输出系统 (BIOS)，它存储在只读存储器芯片上。这个想法可以追溯到加里·基尔多尔 (Gary Kildall) 的 CP/M：让 BIOS 成为唯一可以容纳根据特定机器的具体情况定制操作系统的代码的地方。 IBM 拥有个人计算机 BIOS 中的代码，并起诉任何未经许可使用该代码的公司。

The personal computer revolution seems to have little to do with the age of mainframes that preceded it, but with the passage of time, we can find common themes. IBM’s success with its System/360, and its need to give out a lot of technical information about it, led to the plug compatible industry, which in turn led to IBM’s having to adjust its own product line. Something similar happened with the PC, only this time with a different outcome. Most of the IBM PCs, including the 8088 microprocessor, consisted of parts made by other manufacturers, who were free to sell those parts elsewhere. Microsoft, for instance, retained the right to sell its operating system to others. The core of what made a personal computer an “IBM PC” was the basic input-output system (BIOS), which was stored on a read-only memory chip. The idea went back to Gary Kildall’s CP/M: let the BIOS be the only place where there could be code that tailored the operating system to the specifics of a particular machine. IBM owned the code in the personal computer’s BIOS and prosecuted any company that used it without permission.

在 PC 发布前后，德州仪器的三名员工正考虑辞职，创办一家自己的公司，他们称之为 Compaq。传说 Rod Canion、Jim Harris 和 Bill Murto 在休斯顿一家餐馆的餐巾纸上勾勒出了一台兼容 IBM 的 PC。他们构思了对 IBM PC 进行逆向工程并生产一台 100% 兼容的机器的想法。为了绕过 IBM 对 BIOS 代码的所有权，他们雇佣了对该代码一无所知的人，将他们安置在一个“洁净室”中，在那里他们不会被任何偷偷给他们的禁忌代码所影响，并让他们自己设计一个 BIOS 来复制 IBM 的功能。这虽然很昂贵，但却是合法的。1983 年推出的 Compaq 电脑虽然很重，但便于携带。这实际上是一种营销策略：他们以 25 磅的重量“赋予了举重一词新的含义”。它成功的原因在于它以具有竞争力的价格与 IBM PC 完全兼容。到 1985 年，康柏的销量已使该公司跻身计算机公司 100 强之列，这是所有初创公司中发展最快的公司之一^。83

Around the time of the PC’s announcement, three Texas Instruments employees were thinking of leaving their jobs and starting a company of their own, which they called Compaq. Legend has it that Rod Canion, Jim Harris, and Bill Murto sketched out an IBM-compatible PC on a napkin in a Houston restaurant. They conceived of the idea of reverse-engineering the IBM PC and producing a machine that would be 100 percent compatible. To get around IBM’s ownership of the BIOS code, they hired people who had no knowledge of that code, put them in a “clean room,” where they would not be corrupted by anyone sneaking the forbidden code to them, and had them come up with a BIOS of their own that replicated the functions of IBM’s. This was expensive, but it was legal. The Compaq computer, delivered in 1983, was portable, although heavy. That was really a marketing ploy: At twenty-five pounds they “gave new meaning to the phrase pumping iron.” What made it a success was its complete compatibility with the IBM PC at a competitive price. Compaq’s sales propelled the company into the top 100 rankings of computer companies by 1985, one of the fastest trajectories of any start up.⁸³

康柏公司为打破 IBM 对其 PC 架构的控制而做出的英勇努力并非一蹴而就。一家名为 Phoenix Technologies 的小公司也对 BIOS 芯片进行了逆向工程，他们没有围绕它制造计算机，而是直接出售 BIOS 芯片。现在，制造一台 IBM 兼容 PC 变得容易了。行业媒体发起了一项兼容性测试：这台机器能否运行 Lotus 1-2-3，该软件旨在利用 PC 的内部工作原理来获得最大速度？更妙的是，它能否运行 Flight Simulator，这是 Bruce Artwick 编写的一款程序，可以充分发挥 IBM 架构的每一个细节？⁸⁴如果答案是肯定的，那么这台机器就是真正的克隆机。闸门打开了。与 System/360 和即插即用兼容机时代的成功举措不同，这一次 IBM 失去了对自己架构的控制。

Compaq’s heroic efforts to break through IBM’s control of its PC architecture did not have to be repeated too often. A small company named Phoenix Technologies also reverse-engineered the BIOS chip, and instead of building a computer around it, they simply offered a BIOS chip for sale. Now building an IBM-compatible PC was easy. The trade press instituted a test for compatibility: would the machine run Lotus 1-2-3, which was written to take advantage of the PC’s inner workings to gain maximum speed? Better still, would it run Flight Simulator, a program written by Bruce Artwick that exercised every nook and cranny of the IBM architecture?⁸⁴ If the answer was Yes and Yes, the machine was a true clone. The floodgates opened. Unlike its successful footwork during the times of System/360 and the plug compatibles, this time IBM lost control over its own architecture.

IBM 兼容机和 Macintosh 的推出标志着个人计算开创阶段的结束。小型计算机和大型计算机制造商再也无法忽视这一现象。在 20 世纪 80 年代末，Novell 等公司为个人计算机引入了更强大的网络功能，这使得 PC 网络能够对许多大型系统构成严重挑战。在基于 8 位设计的犹豫不决的开端之后，制造商开发出了与台式机兼容的便携式计算机（图 8.8、8.9）。在微软创造的市场不断推动下，商业软件催生出同样挑战小型计算机和大型计算机领域的应用程序。到 1991 年，基于英特尔 8086 芯片的高级版本并运行 Windows 3.1 的 IBM 兼容计算机将 Macintosh 的功能带入了商业和商业领域。由于 IBM 管理不善等原因，康柏和戴尔等公司通过销售 IBM 兼容计算机获得的利润比 IBM 多。 IBM 仍然是主要的供应商，但最大的赢家是微软，其操作系统与 IBM 计算机及其克隆产品一起销售。

The introduction of IBM Compatibles and the Macintosh signaled the end of the pioneering phase of personal computing. Minicomputer and mainframe manufacturers could no longer ignore this phenomenon. In the late 1980s, companies like Novell would introduce more capable networking abilities for personal computers, which allowed networks of PCs to seriously challenge many large systems. After some hesitant beginnings based on 8-bit designs, manufacturers developed portable computers that were compatible with those on the desktop (figs. 8.8, 8.9). Commercial software, driven relentlessly by the marketplace created by Microsoft, led to applications that likewise challenged the mini and mainframe world. By 1991 the IBM-compatible computers, based on advanced versions of the Intel 8086 chip and running Windows 3.1, brought the Macintosh’s features to the business and commercial world. For reasons having to do more with IBM’s poor management than anything else, companies like Compaq and Dell would earn more profits selling IBM-compatible computers than IBM would. IBM remained a major vendor, but the biggest winner was Microsoft, whose operating system was sold with both IBM computers and their clones.

图 8.8

早期的“便携式”计算机。奥斯本，约 1981 年。除了体积小巧之外，这款计算机还配备了 CP/M 操作系统和应用软件，售价不到 2,000 美元，同样具有革命性。

Figure 8.8

An early “transportable” computer. Osborne, ca. 1981. Just as revolutionary as its small size was the fact that the computer came with the CP/M operating system and applications software, all for less than $2,000.

图 8.9

早期的“膝上型”计算机。Tandy Radio Shack TRS-80，型号 100，约 1983 年。与 Osborne 一样，它使用 8 位微处理器。系统软件和 BASIC 编程语言由 Microsoft 提供，并包含在机器中。此处展示的机器经过多次修改和扩展，多年来一直是作者的家用电脑。（来源：史密森学会。）

Figure 8.9

An early “laptop” computer. Tandy Radio Shack TRS-80, Model 100, ca. 1983. Like the Osborne, it used an 8-bit microprocessor. System software and the BASIC programming language were supplied by Microsoft and included with the machine. The machine shown here was much modified and extended and served as the author’s home computer for many years. (Source: Smithsonian Institution.)

个人电脑彻底改变了办公环境，但它并没有成为政治或文化意义上的革命性机器，而这正是斯图尔特·布兰德等人所预测和希望的。电脑“走进了人民”，但代价是：企业控制。

The personal computer revolutionized the office environment, but it had not become a revolutionary machine in the political or cultural sense, the sense that Stewart Brand and others had predicted and hoped for. Computers came “to the people,” but for a price: corporate control.

工作站、UNIX 和网络，1981 年至 1995 年

Workstations, UNIX, and the Net, 1981–1995

VAX 将科学大型机的功能带入了公司的工程部门。从 20 世纪 80 年代开始，一种新型计算机将这种功能带入了个人桌面。这些“工作站”通过使用廉价的微处理器（通常是摩托罗拉 68000）来实现这一点。成本较低是相对的，低于 VAX 但远高于 PC。它们的架构和物理设计也与个人电脑有很多共同之处。不同之处在于它们使用 UNIX 操作系统，并且具有广泛的网络功能，可以共享数据和昂贵的外围设备（如绘图仪）。

The VAX brought the power of a scientific mainframe into the engineering division of a company. Beginning in the 1980s a new class of computers brought that power to the individual desktop. These “workstations” did that by using an inexpensive microprocessor, typically the Motorola 68000. The lower cost was relative, less than a VAX but much more than a PC. Their architecture and physical design also had much in common with personal computers. The difference was their use of the UNIX operating system, and their extensive networking abilities that allowed sharing data and expensive peripherals like plotters.

第一个走出大门的是马萨诸塞州切姆斯福德的阿波罗公司。其创始人比尔·波杜斯卡 (Bill Poduska) 之前曾与他人共同创办了 Prime，该公司率先推出了 32 位迷你电脑。1981 年，阿波罗公司推出了一款使用摩托罗拉微处理器及其自己的操作系统和网络系统的产品，名为 Domain。1一台工作站（这个名字显然是当时的起源）的起价为 40,000 美元。2^{正如王和施乐已经发现的}^那样，在每个员工的办公桌上配备一台与其他机器联网的计算机比通过“哑”终端访问的集中式分时计算机效率更高。这些工作站卖得很好，被 GE-Calma 和 Mentor Graphics 等公司采用，他们用它们进行电路板、飞机部件和汽车等产品的计算机辅助设计和工程。到 1980 年中期，阿波罗公司已经出货了 1,000 套系统。它很快就遇到了竞争，并于 1989 年被惠普收购，而惠普于 1985 年凭借自己设计的工作站 (9000) 进入市场^。3

First out of the gate was Apollo, of Chelmsford, Massachusetts. Its founder, Bill Poduska, had previously cofounded Prime, the company that pioneered the 32-bit mini. In 1981 Apollo delivered a product that used the Motorola microprocessor and its own operating and networking systems, called Domain.¹ The price for a single workstation (the name apparently originated at this time) began at $40,000.² As Wang and Xerox had already discovered, having a computer at each worker’s desk, networked to other machines, was more efficient than having a centralized time-shared computer accessed through “dumb” terminals. The workstations sold well to firms like GE-Calma and Mentor Graphics, who used them for computer-aided design and engineering of products like circuit boards, aircraft components, and automobiles. By mid-1980 Apollo had shipped 1,000 systems. It soon encountered competition, and in 1989 it was acquired by Hewlett-Packard, which had entered the market with a workstation (the 9000) of its own design in 1985.³

很快，一家位于硅谷、距离苹果公司不远的新公司也加入了竞争。SUN 微系统公司由维诺德·科斯拉于 1982 年初创立，该公司延续了通过调动关键人员将公共资助大学研究项目技术转移到盈利性公司的悠久传统。这个项目是斯坦福大学网络工作站（公司因此得名）；调动人员是斯坦福大学的 Andy Bechtolsheim。同时调动的还有软件；Bill Joy 在伯克利大学就读期间，利用 ARPA 的资助改进了 UNIX 操作系统（图 9.1）。Joy 于 1982 年 6 月跨越海湾加入 SUN 公司^。4 SUN 公司已于 5 月推出了一款工作站，随后不久又推出了功能更强大的 SUN-2。价格在 20,000 美元左右。Bill Joy 带到山景城的东西中就有伯克利 UNIX。

Competition soon came from a new company located just down the road from Apple in Silicon Valley. SUN Microsystems, founded in early 1982 by Vinod Khosla, continued the long tradition of effecting a transfer of technology from a publicly funded university research project to a profit-making company by moving key people. In this case the project was the Stanford University Networked workstation (hence the company’s name); the person was Andy Bechtolsheim of Stanford. A parallel transfer brought its software; Bill Joy, who, with ARPA funding, had enhanced the UNIX operating system while at Berkeley (figure 9.1). Joy moved across the Bay to join SUN in June 1982.⁴ SUN had already introduced a workstation in May, with the more capable SUN-2 following shortly. Prices were in the $20,000 range. Among the things that Bill Joy brought with him to Mountain View was Berkeley UNIX.

UNIX：从新泽西到加利福尼亚

UNIX: From New Jersey to California

比尔·乔伊转投 SUN 标志着 UNIX 横跨大陆之旅的最后一站：它始于新泽西，在香槟-厄巴纳停留了一段时间，并在伯克利进行了大规模改写。在硅谷，它从学术领域转向商业用途。伯克利 UNIX 是 SUN 成功的关键，并在 20 世纪 90 年代帮助互联网摆脱了 ARPA 的束缚。

Bill Joy’s move to SUN signified the last stop on a transcontinental journey for UNIX: it began in New Jersey, stopped in Champaign-Urbana for a while, and was extensively rewritten while at Berkeley. In Silicon Valley it would move from its academic niche into commercial use. Berkeley UNIX was a key to SUN’s success and helped push the Internet out of its ARPA roots in the 1990s.

UNIX 诞生于贝尔实验室，该实验室隶属于受监管的垄断企业 AT&T。在 1981 年解散之前，AT&T 已同意不参与商业计算活动；作为回报，它从全美提供电话服务的业务中获得了稳定而定期的利润。肯·汤普森和丹尼斯·里奇说，他们最初认为 UNIX 只供贝尔实验室使用，但几个因素共同作用，几乎确保了它“逃脱”，⁵最重要的是，AT&T 不会为了盈利而出售它。因此，大学可以以象征性的成本获得 UNIX 许可证——最多几百美元（商业客户必须支付更多）。同样重要的是，UNIX 并不是当时人们所理解的完整操作系统，而是一组允许用户以简单直接的方式操作文件的基本工具。

Bell Laboratories, where UNIX was created, was a part of AT&T, a regulated monopoly. Before it breakup in 1981, AT&T had agreed not to engage in commercial computing activities; in return it enjoyed steady and regular profits from its business of providing telephone service throughout the United States. Ken Thompson and Dennis Ritchie said that they initially thought of UNIX as something to be used only within Bell Labs, but several factors conspired to all but ensure that it would “escape,”⁵ most importantly, that AT&T would not offer to sell it for a profit. Thus universities could obtain a UNIX license for a nominal cost— a few hundred dollars at most (commercial customers had to pay more). Also important was that UNIX was not a complete operating system, as it was then understood, but rather a set of basic tools that allowed users to manipulate files in a simple and straightforward manner.

图 9.1

Bill Joy。（来源：SUN Microsystems。）

Figure 9.1

Bill Joy. (Source : SUN Microsystems.)

结果，UNIX 成了大学计算机科学系的天赐之物。AT&T 的 Western Electric 子公司只需支付象征性的费用，即可提供 UNIX 的源代码。代码是用 C 编程语言编写的，而不是机器语言。这意味着，尽管 UNIX 是在 DEC 计算机上开发的，但它可以在任何带有 C 编译器的机器上运行。相比之下，大多数计算机供应商将源代码视为传家宝，很少将其泄露，并尽其所能将客户锁定在其产品中。而且，没有人介意大学修改 UNIX 以增强其功能。这正是研究生（以及许多聪明的本科生）在校学习的目的。因此，将 AT&T 的 UNIX 变成实用系统所需的所有工作（例如，针对特定的显示器、打印机和存储系统进行定制）都可以以低成本完成。这项工作对学生来说也是极好的培训。当这些学生毕业时，他们带着这些技能，以及对给予他们如此自由的软件的欣赏。其中一些人在毕业并进入“现实世界”后发现，公司计算机中心几乎没有空间提供这种轻松访问机器底层的功能。这些程序员将自己变成了福音传道者，将 UNIX 编程方式传播到企业界。

The result was that UNIX was a godsend for university computer science departments. For a nominal fee, AT&T’s Western Electric subsidiary supplied UNIX’s source code. The code was written in the C programming language, not machine language. That meant that although developed on DEC computers, UNIX could run on any machine that had a C compiler. By contrast, most computer vendors guarded source code as their family jewels, seldom gave it out, and did all they could to lock a customer into their products. And no one minded if a university modified the UNIX to enhance its capabilities. That was what graduate students—and many bright undergraduates as well—were in school for. Thus all the things needed to turn AT&T’s UNIX into a practical system—for example, tailoring it for specific monitors, printers, and storage systems—got done cheaply. That work was excellent training for students as well. When these students graduated, they took these skills with them, along with an appreciation for the software that had allowed them such freedom. Some of them found after graduation and entry into the “real world” that corporate computer centers had little room for that kind of easy access to the lower levels of a machine. Those programmers turned themselves into evangelists, spreading the UNIX way of programming into the corporate world.

比尔·乔伊 (Bill Joy) 是众多对 AT&T 版本的 UNIX 进行修改并希望改进的学生之一。1974 年，在肯·汤普森 (Ken Thompson) 到访后，加州大学伯克利分校获得了一盘 UNIX 磁带。该系统很快就在校园里的几台 PDP-11 上运行起来。比尔·乔伊也于同年来到校园^。6

Bill Joy was one of many students who had tinkered with AT&T’s version of UNIX hoping to make it better. The University of California at Berkeley obtained a UNIX tape in 1974, following a visit by Ken Thompson. The system was soon running on several PDP-11s on the campus. Bill Joy also arrived on the campus that year.⁶

汤普森和里奇创建 UNIX 的直接目标是找到一种轻松共享文件的方法。在实验室于 1969 年取消了 Multics 的工作后，他们还需要编程工具。在 PDP-7 上完成初始工作后，他们将 UNIX 移至 PDP-11 并用 C 语言重写了它。由于这些原因，UNIX 极其简洁：大多数命令都使用两三个字母的缩写。有时这些缩写与命令的作用相对应（例如“cp”表示复制）；其他时候这种关系很脆弱（例如“grep”：全局搜索正则表达式并打印）。7 ^UNIX的原则之一是任何 UNIX 进程的输出都可以用作另一个进程的输入。这赋予了 UNIX 巨大的能力和灵活性。这也意味着缺少分页符或显示正在处理的文件状态的信息等功能，因为如果“传输”到另一个进程，这些会使文件变得混乱。它还使编写像“病毒”一样的程序变得容易，这些程序可以通过生成可执行代码作为输出来自我复制。UNIX 功能强大，但对普通人来说却没什么用^。8

Thompson and Ritchie’s immediate goal in creating UNIX was to have a way of sharing files easily. They also were in need of programming tools, after the Labs canceled work on Multics in 1969. After the initial effort on the PDP-7, they moved UNIX to a PDP-11 and rewrote it in C. For those reasons UNIX was frugal in the extreme: two-or three-letter abbreviations were the norm for most commands. Sometimes these abbreviations corresponded to what the command did (e.g. “cp” for copy); other times the relationship was tenuous (e.g. “grep”: globally search for the regular expression and print).⁷ One of UNIX’s tenets was that the output of any UNIX process be usable as input for another. That gave UNIX enormous power and flexibility. It also meant an absence of features like page-breaks or information that revealed the status of the file being worked on, since these would clutter up the file if “piped” to another process. It also made it easy to write programs that acted like a “virus,” programs that could replicate themselves by producing executable code as output. UNIX was powerful, but not useful for the hoi polloi.⁸

比尔·乔伊 (Bill Joy) 和他在伯克利的同学们开始着手让 UNIX 更易于访问。最初的动力来自于原始的 Model 33 电传打字机被“愚蠢的”基于 CRT 的终端 (Lear-Siegler ADM-3s) 所取代。到 1978 年，乔伊以象征性的价格向全国各地的朋友和同事提供第一版伯克利软件发行版 (BSD) 的磁带。9^贝尔实验室 UNIX 的增强功能使 PDP-11 的功能不堪重负，工作转移到 DEC 刚刚推出的 VAX 上。1980 年，ARPA 支持伯克利 UNIX，将其作为该机构可以向所有客户推荐的通用系统。从理论上讲，UNIX 可以移植到 DEC 以外的制造商的计算机上，这是主要原因。在伯克利 UNIX（4.2 BSD 版）中添加的众多增强功能中，有一项是支持通过称为 TCP=IP 的协议进行联网，ARPA 将其推广为一种互连网络的方式。该协议及其与 Berkeley UNIX 的捆绑永远将 UNIX 与互联网联系在一起^。10

Bill Joy and his fellow students at Berkeley set out to make UNIX more accessible. The initial impetus came when the primitive Model 33 Teletypes were replaced by “dumb” CRT-based terminals (Lear-Siegler ADM-3s). By 1978 Joy was offering tapes of the first Berkeley Software Distribution (BSD) at a nominal cost to his friends and colleagues around the country.⁹ The enhancements to Bell Labs’s UNIX strained the capabilities of the PDP-11, and work shifted over to a VAX, which DEC had just introduced. In 1980 ARPA threw its support behind Berkeley UNIX as a common system the agency could recommend for all its clients. That UNIX was, in theory, portable to computers from manufacturers other than DEC was a main reason. Among the many enhancements added to Berkeley UNIX (in version 4.2 BSD) was support for networking by a protocol known as TCP=IP, which ARPA promoted as a way to interconnect networks. This protocol, and its bundling with Berkeley UNIX, forever linked UNIX and the Internet.¹⁰

UNIX 的讽刺之处

The Ironies of UNIX

尽管 UNIX 是 Thompson 和 Ritchie 为自己和像他们这样的研究人员编写的，但它还是得到了普遍使用。UNIX 的优势在于 Thompson 和 Ritchie 对自己想要什么和不想要什么有着坚定的认识；然而，它确实产生了许多不兼容、复杂且功能繁琐的版本。Berkeley UNIX 是在 VAX 上开发的；Digital Equipment 只是勉强容忍 UNIX 在 VAX 上运行，并将其客户引导到 VMS。运行 Berkeley UNIX 和 TCP=IP 的 VAX 计算机帮助将受限的 ARPANET 转变为广阔的互联网；当互联网在 1990 年代开始向公众使用时，Digital Equipment Corporation 几乎没有获利。UNIX 诞生于大学环境中，以简化文件共享的方式而闻名；这也意味着 UNIX 系统容易受到病毒和黑客未经授权的入侵。UNIX 之所以传播开来，是因为 AT&T 免费提供了它；伯克利 UNIX 为 SUN 和其他工作站供应商带来了巨额利润，但 AT&T 却没有。当 AT&T 在被剥离后获准销售计算机产品时，它未能为其创造的产品制定标准或从中获利。AT&T 吹捧 UNIX 是一个“开放”系统；其他公司推出不兼容的版本，仅仅是为了与竞争对手 AT&T 有所区别。UNIX 这个名字是 Multics 的双关语，意为一个更简单的系统；它并不代表统一，而且在 1985 年之后，它就不再具有通用性了。最后，尽管 UNIX 具有诸多优点，但它从未挑战过微软的 MSDOS=Windows 的主导地位，后者在 1981 年后成为真正的桌面标准。这就是 UNIX 的讽刺之处。

Although UNIX was written by Thompson and Ritchie for themselves and for researchers like them, it found its way into general use. UNIX’s strength came from the fact that Thompson and Ritchie had a firm sense of what they wanted and what they did not want; it did, however, sprout a number of incompatible, baroque, and feature-laden versions. Berkeley UNIX was developed on a VAX; Digital Equipment only grudgingly tolerated UNIX on the VAX and steered its customers to VMS. VAX computers, running Berkeley UNIX with TCP=IP, helped transform the restricted ARPANET to the wide-open Internet; when the Internet broke through to public use in the 1990s, Digital Equipment Corporation hardly profited. UNIX, born in a collegial environment, was best known for the way it made the sharing of files easy; that also meant that UNIX systems would be vulnerable to viruses and unauthorized intrusions by hackers. UNIX spread because AT&T gave it away; Berkeley UNIX generated enormous profits for SUN and other workstation vendors but not for AT&T. When AT&T was allowed to market computer products after divestiture, it failed to set a standard for, or profit from, its creation. AT&T touted UNIX as an “open” system; other companies introduced incompatible versions for little more reason than to be different from AT&T, a competitor. The name UNIX was a pun on Multics, chosen to imply a simpler system; it was not intended to imply unity, and after 1985 it was anything but universal. Finally, for all its qualities, UNIX never challenged the dominance of Microsoft’s MSDOS=Windows, which became the real desktop standard after 1981. Such were the ironies of UNIX.

SUN Microsystems 充分利用了开放系统战略。它不仅使用了 UNIX 和摩托罗拉微处理器，还使用了标准化的以太网版本和其他人可以自由采用的内部总线。阿波罗和惠普保留了他们的专有系统，尽管它们最终也提供了 UNIX。SUN 模式、利润和风险资本的可用性催生了 JAWS 现象（只是另一个工作站）。许多竞争对手试图在特定的领域找到立足点：例如，Symbolics 生产了一种工作站，该工作站经过优化，可运行用于人工智能应用的 LISP 编程语言。但大多数竞争对手在 SUN 的开放通用机器面前失败了，这些机器以低廉的价格提供良好的性能。除了惠普和阿波罗之外，唯一的例外是 Silicon Graphics。与 SUN 一样，Silicon Graphics 将一个大学赞助的研究项目商业化；一个“几何引擎”芯片，可执行表示三维图像所需的计算。和 SUN 一样，技术转让也通过调动一位关键人物来实现——吉姆·克拉克 (Jim Clark)，他曾在犹他州师从大卫·埃文斯 (David Evans) 和伊万·萨瑟兰 (Ivan Sutherland)，并在斯坦福大学开发了原型引擎。（克拉克后来离开 Silicon Graphics，创办了 Netscape，一家致力于将互联网软件商业化的公司。）

SUN Microsystems took full advantage of the strategy of open systems. It not only used UNIX and the Motorola microprocessor but also a standardized version of Ethernet, and an internal bus that others were free to adopt. Apollo and Hewlett-Packard retained their proprietary systems, although each eventually offered UNIX as well. The SUN model, its profits, and the availability of venture capital spawned the JAWS phenomenon (just another work station). Many of the competitors tried to find a toehold in a specific niche: for example, Symbolics produced a workstation that was optimized to run the LISP programming language for artificial intelligence applications. But most failed in the face of SUN’s open, general-purpose machines offering good performance at a low price. Besides HP and Apollo, the only serious exception was Silicon Graphics. Like SUN, Silicon Graphics commercialized a university-sponsored research project; a “geometry engine” chip that performed the calculations needed to represent three-dimensional images. Like SUN, technology transfer was effected by moving a key person—Jim Clark, who had studied under David Evans and Ivan Sutherland at Utah and had developed a prototype engine at Stanford. (Clark later left Silicon Graphics to found Netscape, a company aimed at commercializing software for the Internet.)

VAX 策略

VAX Strategy

正如个人计算机领域分为DOS和Macintosh两大阵营一样，在科学和工程领域也进行了一场战争。

Just as the personal computer field was divided into the DOS and Macintosh camps, there was also a battle going on in the scientific and engineering field.

工作站公司无法在单台机器的性能上与小型机和大型机竞争；他们通过销售机器网络来竞争，他们声称这些机器的总体性能大于各部件的总和。SUN 公司在其广告口号中简明扼要地表达了这一点，“网络就是计算机。”整个 80 年代，数字设备公司都有自己的强大战略，将 SUN 公司对网络的重视与 IBM 的统一计算机系列概念结合起来。DEC 的计划是向客户提供单一架构 VAX，使用单一操作系统 VMS，提供从台式机到大型机功能的单独或联网配置。VAX 战略中唯一不属于数字设备公司自己的部分是网络 — — 以太网，DEC 通过与英特尔和施乐的协议获得了以太网。VAX 11=780 之后是更小的机器，例如 1980 年的 11=750 和 1984 年的 MicroVax II；以及更大型的机器，如 1984 年的 8600（Venus）和 1990 年的 9000（DEC 称之为大型机）^。11

Workstation companies could not compete with mini and mainframes on the basis of the power of a single machine; they competed by selling networks of machines, whose collective power they alleged was greater than the sum of the parts. SUN stated this succinctly in its advertising slogan, “The network is the computer.” Throughout the 1980s Digital Equipment Corporation had a powerful strategy of its own that combined SUN’s emphasis on networking with IBM’s concept of a unified family of computers. DEC’s plan was to offer the customer a single architecture, the VAX, with a single operating system, VMS, in solitary or networked configurations that ranged from desktop to mainframe capability. The only part of the VAX Strategy that was not Digital’s own was the networking—Ethernet, which DEC obtained in an agreement with Intel and Xerox. The VAX 11=780 was followed by smaller machines like the 11=750 in 1980 and MicroVax II in 1984; and larger machines like the 8600 (Venus) in 1984 and the 9000 (which DEC called a mainframe) in 1990.¹¹

VAX 战略本身也有风险，类似于 IBM 用 System=360 “押上整个公司”的风险。DEC 必须让客户相信，它可以提供从办公自动化软件到打印机色带等所有产品，但又不能让客户觉得它的价格过高。它必须设计和制造性能良好的产品。DEC 必须停止销售自己的竞争架构，包括 PDP-10 衍生的 36 位计算机。PDP-10 系列基于旧设计，与 VAX 不兼容；将 PDP-10 应用程序简单地转换为 VAX 似乎很简单。

The VAX Strategy had risks of its own, risks that resembled IBM’s “betting the company” with the System=360. DEC had to convince the customer that it could supply everything, from office automation software to printer ribbons, yet not convey a sense that it was charging excessively high prices. It had to design and build products with good performance across the entire line. DEC had to stop marketing its own competing architectures, including the 36-bit computers descended from the PDP-10. The PDP-10 series was based on an old design and was incompatible with the VAX; a simple conversion of PDP-10 applications to the VAX seemed straightforward.

DEC 错误地估计了 PDP-10 在其客户中的受欢迎程度——他们要么忘记了，要么从未意识到，现代计算机在多大程度上是这台古老机器的产物。当 DEC 宣布逐步淘汰 DECtape 时，甚至引发了强烈抗议。DEC 宣布 1983 年以后不再开发 PDP-10 机器，这一决定遭到了客户的强烈抵制，但 Ken Olsen 态度坚定。这是一个理性的决定吗？是的，但如何衡量它的无形资产？PDP-10 是第一个创造个人计算幻觉的系统。它的 TOPS-10 操作系统启发了个人计算机系统软件。威廉·盖茨就是在这台计算机上学习编程，并编写 Microsoft BASIC。早期的 ARPANET 连接了比其他任何计算机都多的 PDP-10。PDP-10 硬件早已过时。它所创造的计算思维模型不仅仍然存在，而且每次我们打开联网的个人计算机或工作站时，它都会出现。

DEC misjudged how beloved the PDP-10 was among its customers— either forgetting, or perhaps never realizing, how much modern computing was a child of that venerable machine. There was even an outcry when DEC announced it was phasing out DECtape. DEC’s announcement that no PDP-10 machines would be developed after 1983 was met by strong customer resistance, but Ken Olsen stood firm. A rational decision? Yes, but how does one measure its intangibles? The PDP-10 was the system that first created the illusion of personal computing. Its TOPS-10 operating system inspired personal computer system software. It was the computer that William Gates learned to program on, and to write Microsoft BASIC on. The early ARPANET linked up more PDP-10s than any other computer. The PDP-10 hardware was long obsolete. The mental model of computing that it created is not only still alive, it greets us every time we turn on a networked personal computer or workstation.

但在个人电脑层面，Vax 策略发生了偏离。1982 年，DEC 推出了三台不兼容的机器，而不是一台。这些机器不仅与 VAX 不兼容，而且与 IBM PC 也不完全兼容。其中一台 Rainbow 销量不高，但无法阻挡 IBM 兼容机的崛起。12^与IBM PC 标准不完全兼容是一个致命错误。

But there was a deviation from the VAX strategy—at the personal computer level. There DEC introduced not one but three incompatible machines in 1982. Not only were these incompatible with the VAX, they were not fully compatible with the IBM PC either. One of them, the Rainbow, sold modestly but could not slow down the IBM-compatible juggernaut.¹² The lack of full compatibility with the IBM PC standard was a fatal error.

VAX 策略在整个 20 世纪 80 年代都发挥了良好的作用。到那时，IBM 360-370 架构已变得过于复杂，需要进行增强、修改和扩展。与此同时，IBM 的中端 System=38 和 PC 销售火爆，但两者都与 System=370 不兼容。IBM 的客户感到困惑。IBM 销售人员从小就销售“大型机”（大型主机），将任何小型机都视为玩具，他们经常试图引导每位客户购买 System=370，这让客户更加困惑。在 1987 年 10 月股市崩盘之前的一个短暂而辉煌的时刻，DEC 似乎不仅稳居第二，而且还准备做出不可思议的事情，超越 IBM 并主宰行业。但事实并非如此。 DEC 的股票是当月跌幅最大的股票之一，到 1990 年，由于其 VAX 战略的缺陷，加上无法将新的 VAX 产品推向市场，DEC 开始在一系列季度中损失了整个小型计算机时代赚到的大部分钱。如果 UNIX 工作站或 IBM PC 单独出现，DEC 可能可以经受住攻击，但两者结合起来就太厉害了。

The VAX Strategy worked well through the 1980s. By that time the IBM 360-370 architecture was becoming top-heavy with enhancements, modifications, and extensions. Meanwhile IBM was enjoying brisk sales of its mid-range System=38 and of course its PC, neither compatible with the System=370. IBM’s customers were confused. IBM salesmen, who grew up selling “big iron” (large mainframes) and who regarded anything smaller as toys, often added to the confusion by trying to steer every customer to the System=370. For a brief and glorious moment—just before the stock market crash in October 1987—it looked as if DEC was not only in a solid number two position but poised to do the unthinkable, surpass IBM and dominate the industry. But it was not to be. DEC’s stock was among the heaviest losers that month, and by 1990 the drawbacks to its VAX strategy, combined with its inability to bring new VAX products to market, began a series of quarters in which DEC lost most of the money it had earned through the entire minicomputer era. DEC probably could have weathered an assault from UNIX workstations or from the IBM PC if either occurred alone, but the combination was too much.

精简指令集

RISC

如果说这两个错误还不够，DEC 还犯了第三个错误。这个错误涉及到该公司最强项：计算机架构。简而言之，DEC 未能开发出新的架构来反映 20 世纪 70 年代芯片和软件技术的变化。

If those two blunders were not enough, DEC made a third. This blunder involved what had been the company’s strongest suit; computer architecture. Simply put, DEC failed to develop new architectures to reflect the changes in chip and software technology that had taken place in the 1970s.

尽管 VAX 的名称暗示了 PDP-11 小型计算机的扩展，但它的架构与 IBM System=360 及其后代有很多共同之处。与 360 一样，它的指令集包含在一个微程序中，存储在只读存储器中。与 360 一样，VAX 为程序员提供了一套丰富的指令，这些指令几乎可以以所有可以想象的方式对数据进行操作。370=168 有超过 200 条指令，VAX 11=780 有超过 250 条指令。有针对整数、浮点数、压缩十进制数和字符串的指令集，以多种模式运行。13^这种理念是在磁芯存储器占主导地位的环境中发展起来的，相对于处理器操作而言，访问磁芯存储器的速度较慢。因此，在进入内存获取数据之前，详细说明想要对数据执行的操作是有意义的。指令集还反映了编译器技术的状态。如果处理器能够仅用一条指令对数据执行大量算术运算，那么编译器的工作量就会少很多。丰富的指令集将减少高级编程语言的类似英语的命令与机器代码的原始和繁琐命令之间的“语义鸿沟”。廉价的只读存储器芯片意味着，如果计算机是微编程的，设计人员可以以低成本创建这些丰富的指令集^。14

Although its name implied an extension of the PDP-11 minicomputer, the VAX architecture had a lot in common with the IBM System=360 and its descendants. Like the 360, its instruction set was contained in a microprogram, stored in a read-only memory. Like the 360, the VAX presented its programmers with a rich set of instructions that operated on data in almost every conceivable way. The 370=168 had over 200 instructions, the VAX 11=780 over 250. There were sets of instructions for integers, floating-point numbers, packed decimal numbers, and character strings, operating in a variety of modes.¹³ This philosophy had evolved in an environment dominated by magnetic core memory, to which access was slow relative to processor operations. Thus it made sense to specify in great detail what one wanted to do with a piece of data before going off to memory to get it. The instruction sets also reflected the state of compiler technology. If the processor could perform a lot of arithmetic on data with only one instruction, then the compiler would have that much less work to do. A rich instruction set would reduce the “semantic gap” between the English-like commands of a high-level programming language and the primitive and tedious commands of machine code. Cheap read-only memory chips meant that designers could create these rich instruction sets at a low cost if the computer was microprogrammed.¹⁴

这些假设早已被接受。但计算机科学并非停滞不前。20 世纪 70 年代中期，IBM 的 John Cocke 关注编译器的快速发展，并得出结论：使用更频繁的命令将数据加载到内存和从内存存储数据，一组较小的指令可以比 System=370 运行得更快。Thomas Watson Jr. 曾写过一份备忘录，描述了 IBM 需要在其队伍中拥有“野鸭”——那些不满足于接受传统观念的人。在 IBM 保守的文化中培养这样的人才并不容易，但 Watson 比其他任何计算机高管都更清楚，没有他们，IBM 就无法生存。John Cocke 当时对计算机设计有着激进的想法，符合这一描述^。15

Those assumptions had been long accepted. But computer science was not stagnant. In the mid-1970s John Cocke of IBM looked at the rapid advances in compilers and concluded that a smaller set of instructions, using more frequent commands to load and store data to and from memory, could operate faster than the System=370. Thomas Watson Jr. once wrote a memo describing IBM’s need to have “wild ducks” among its ranks—-people who were not content to accept conventional wisdom about the way things were done. Cultivating such people in the conservative culture of IBM was not easy, but Watson knew, perhaps better than any other computer executive, that IBM could not survive without them. John Cocke, with his then-radical ideas about computer design, fit that description.¹⁵

Cocke 的想法促成了一台名为 IBM 801 的实验机器的诞生，该机器于 1979 年在 George Radin 的指导下完成。16^由于许多原因，包括 370 系列及其后续产品的成功和利润，IBM 一直没有推出该设计的商业版本。（1986 年推出的 IBM-RT 是一次商业失败，并没有很好地利用这一想法。）尽管如此，801 项目的消息还是传开了，还有传言说它可以以更快的速度执行 System=370 程序，尽管它是一台较小的计算机。到 20 世纪 70 年代末，磁芯已被半导体存储器取代，其访问时间与处理器的速度相匹配。频繁的加载和存储指令不再会造成速度损失。最后，一些研究人员研究了 VAX 并得出结论，他们无法进一步扩展其设计；他们开始寻找替代方案。

Cocke’s ideas led to an experimental machine called the IBM 801, completed under the direction of George Radin in 1979.¹⁶ For many reasons, including the success and profits of the 370 line and its successors, IBM held back introducing a commercial version of the design. (The IBM-RT, introduced in 1986, was a commercial failure and did not exploit the idea very well.) Still, word of the 801 project got out, along with a rumor that it could execute System=370 programs at much faster speeds although it was a smaller computer. By the late 1970s magnetic core had been replaced by semiconductor memory, whose access times matched the speeds of processors. Frequent load and store instructions no longer exacted a speed penalty. Finally, some researchers looked at the VAX and concluded that they could not extend its design any further; they began looking for alternatives.

1980 年，伯克利分校的 David Patterson 领导的一个小组在听到“801 的传闻”后，启动了一个类似的项目，称为 RISC，即“精简指令集计算机”。另一个项目，称为 MIPS（每秒数百万条指令），于 1981 年在斯坦福大学由 John Hennessy 领导启动。17^当他们公开他们的工作时，遭到了质疑：RISC 在大学实验室里看起来不错，但并没有满足实际客户的实际需求（图 9.2）。一份行业杂志甚至担心，从一开始与 UNIX 相关的 RISC 就不适合用 COBOL 编写的数据处理作业。18^与此同时，基于英特尔的个人电脑、VAX 和 System=370 系列（所有复杂指令集处理器）的销售都在蓬勃发展。随着罗纳德·里根总统领导下国防部的大规模扩充，华尔街迎来了又一轮繁荣时期。那些关注 DEC、Data General、IBM 和王安公司股票走势的人并不担心 RISC。

In 1980 a group at Berkeley led by David Patterson, after hearing “rumors of the 801,” started a similar project called RISC—”Reduced Instruction Set Computer.” Another project, called MIPS (Millions of Instructions Per Second), began in 1981 at Stanford under the leadership of John Hennessy.¹⁷ As they publicized their work they were met with skepticism: RISC looked good in a university laboratory but did not address the real needs of actual customers (figure 9.2). One trade journal even worried that RISC, from the start associated with UNIX, was not well-suited for data-processing jobs written in COBOL.¹⁸ Meanwhile, sales of Intel-based PCs, the VAX, and the System=370 family—all complex instruction-set processors—were booming. With a massive buildup of the Defense Department under President Ronald Reagan, Wall Street was enjoying another round of go-go years. Those watching the trajectory of their stocks in DEC, Data General, IBM, and Wang were not worried about RISC.

SUN Microsystems 的产品最初使用的是 Motorola 68000 微处理器，其设计非常符合 PDP-11 和 VAX 的精神。从 1987 年开始，可能由于 Bill Joy 的影响，SUN 推出了一款工作站，该工作站采用基于 Patterson 在伯克利的研究的 RISC 芯片。这种设计被称为 SPARC（可扩展处理器架构），它比其他任何设计都更能打消人们对 RISC 的怀疑。Hennessy 和 Patterson 成为 RISC 的拥护者，他们得到了一些令人印象深刻的定量测量的支持，这些测量表明 RISC 设计可以从硅片中榨取比传统智慧认为可能的更多的处理能力。更能说明问题的是，他们的数据显示，RISC 提供了一种提高微处理器速度的方法，其速度比小型机和大型机的速度提高得更快，或者可以提高得更快。毫无疑问，这意味着，这种诞生于袖珍计算器的廉价微处理器，其性能很快就会超越小型计算机、大型机，甚至超级计算机。如果真是这样，那么他们的结论就意味着，数十年来广为人知的计算机行业（美国司法部在整个 20 世纪 70 年代与 IBM 争夺该行业）正被其后代逼入绝境^。19

SUN Microsystems’ products initially used the Motorola 68000 microprocessor, whose design was very much in the spirit of the PDP-11 and VAX. Beginning in 1987 and probably owing to Bill Joy’s influence, SUN introduced a workstation with a RISC chip based on Patterson’s research at Berkeley. Called SPARC (Scalable Processor Architecture), this design did more than anything else to overcome skepticism about RISC. Hennessy and Patterson became evangelists for RISC, buttressed by some impressive quantitative measurements that showed how a RISC design could squeeze much more processing power out of a piece of silicon than conventional wisdom had thought possible. More telling, their data showed that RISC offered a way of improving microprocessor speeds much more rapidly than mini and mainframe speeds were improving—or could improve. The unmistakable implication was that the puny, cheap microprocessor, born of a pocket calculator, would soon surpass minicomputers, mainframes, and even supercomputers, in performance. If true, their conclusions meant that the computer industry as it had been known for decades, and over which the U.S. Justice Department fought IBM throughout the 1970s, was being driven to extinction by its own offspring.¹⁹

图 9.2

RISC 的优势：微处理器的计算机性能。Hennessy 和 Patterson 使用此类图表来展示廉价微处理器在性能上将很快超越所有其他类型的计算机。（来源：John L. Hennessy 和 Norman P. Jouppi，《计算机技术和架构：不断发展的互动》， IEEE 计算机[1991 年 9 月]：19。）# IEEE。

Figure 9.2

The Benefits of RISC: Computer Performance of Microprocessors. Hennessy and Patterson used graphs such as this one to show how inexpensive microprocessors would soon overtake all other classes of computers in performance. (Source : John L. Hennessy and Norman P. Jouppi, “Computer Technology and Architecture: an Evolving Interaction,” IEEE Computer [September 1991]: 19.) # IEEE.

SUN 更进一步推广 RISC：他们授权 SPARC 设计，以便其他公司可以采用它并使 SPARC 成为标准。20^复制SPARC 处理器的许可证加上 Berkeley UNIX 的结合，使得进入工作站市场几乎和制造 IBM 兼容 PC 一样容易。SUN 赌一把，认为继续推出高性能、低价的产品也会使其受益。他们成功了，尽管这种理念意味着它必须接受微薄的利润率，因为 SUN 不能拥有架构。

SUN went a step further to promote RISC: they licensed the SPARC design so that other companies might adopt it and make SPARC a standard.²⁰ The combination of a license to copy the SPARC processor, plus Berkeley UNIX, made it almost as easy to enter the workstation market as it was to make an IBM compatible PC. SUN gambled that it, too, would benefit by continuing to introduce products with high performance and a low price. They succeeded, although such a philosophy meant it had to accept slim profit margins, since SUN could not own the architecture.

斯坦福 MIPS 项目还催生了一家商业企业 MIPS 计算机系统，这也有助于建立 RISC 微处理器的商业市场。数字设备公司于 1989 年从 MIPS 购买了一块芯片，用于其一台工作站——甚至 DEC 现在也承认 RISC 不会消失。（DEC 内部的 RISC 项目 Prism 已于 1988 年被取消。）Silicon Graphics 也将其较新的工作站基于 MIPS 微处理器。21^惠普将其工作站系列转换为称为精确架构的 RISC 设计。在 RT 失败后，IBM 于 1990 年推出了一款成功的 RISC 工作站，即 R=6000。20 世纪 90 年代初，苹果和 IBM 与摩托罗拉联手生产了一款名为 Power PC 的 RISC 微处理器，他们希望这款微处理器能够颠覆英特尔 8086 系列。 IBM 在 Power PC 设计中所扮演的角色恰如其分地证明了约翰·考克 (John Cocke) 的理念，他被称为“野鸭子”，是这一现象的开创者。

The Stanford MIPS project also spawned a commercial venture, MIPS Computer Systems, which also helped establish a commercial market for RISC microprocessors. Digital Equipment Corporation bought a chip from MIPS for one of their workstations in 1989—even DEC now admitted that RISC was not going away. (An internal RISC project at DEC, called Prism, had been canceled in 1988.) Silicon Graphics also based its newer workstations on MIPS microprocessors.²¹ Hewlett-Packard converted its line of workstations to a RISC design called precision architecture. After failing with the RT, IBM introduced a successful RISC workstation in 1990, the R=6000. In the early 1990s Apple and IBM joined forces with Motorola to produce a RISC microprocessor called Power PC, which they hoped would topple the Intel 8086 family. IBM’s role in the design of the Power PC was a fitting vindication of the ideas of John Cocke, the “wild duck” who started the whole phenomenon.

网络 I：以太网

Networking I: Ethernet

RISC 架构、UNIX 以及科学或工程应用程序将工作站与个人计算机区分开来。另一个区别是，工作站从一开始就设计为联网，尤其是在本地级别，例如在建筑物内或工程公司的某个部门内。这是使用以太网实现的，以太网是施乐帕洛阿尔托研究中心最重要的发明之一。如果说 20 世纪 90 年代的互联网成为“信息高速公路”，那么以太网则成为为其提供信息的同等重要的本地道路网络。作为 ARPA 研究的后代，我们现在称为互联网的全球网络在施乐发明本地以太网之前就已经存在。但以太网在互联网产生重大影响之前就改变了办公和个人计算的性质。因此，我们将首先研究以太网是如何做到这一点的。

A RISC architecture, UNIX, and scientific or engineering applications differentiated workstations from personal computers. Another distinction was that workstations were designed from the start to be networked, especially at the local level, for example, within a building or a division of an engineering company. That was done using Ethernet, one of the most significant of all the inventions that came from the Xerox Palo Alto Research Center. If the Internet of the 1990s became the “Information Superhighway,” then Ethernet became the equally important network of local roads to feed it. As a descendent of ARPA research, the global networks we now call the Internet came into existence before the local Ethernet was invented at Xerox. But Ethernet transformed the nature of office and personal computing before the Internet had a significant effect. How Ethernet did that will therefore be examined first.

赫伯·格罗施 (Herb Grosch) 在自传中自豪地指出，格罗施定律 (见第 6 章) 是在 1950 年提出来的，当时还没有商业化安装，但经过一波又一波的技术创新，它依然屹立不倒。22 ²⁰世纪 70 年代末，人们尤其沮丧地意识到，尽管人们可以在一块 50 美元的芯片上获得 IBM 7090 的所有功能，但购买一组廉价系统所获得的计算能力，不如花同样的钱购买一个大型系统。以太网改变了这一局面，使小型工作站集群以及后来的 PC 能够有效地协同工作。

In his autobiography, Herb Grosch notes with pride that Grosch’s Law (see chapter 6), conceived in 1950 before there were even commercial installations, held through waves of technical innovation.²² In the late 1970s it was especially frustrating to realize that even though one could get all the functions of an IBM 7090 on a fifty-dollar chip, buying an ensemble of cheap systems did not give as much computing power as spending the same money on one large system. Ethernet changed that equation, by enabling small clusters of workstations and, later, PCs to work together effectively.

以太网由罗伯特·梅特卡夫和戴维·博格斯于 1973 年在施乐-PARC 发明。梅特卡夫是麻省理工学院 MAC 项目的校友，1969 年，他帮助将麻省理工学院连接到 ARPANET，并将一台 PDP-10 连接到该网络。1972 年，他转到施乐-PARC；他在那里的首要任务之一就是将 PARC 的 PDP-10 克隆版 MAXC 连接到 ARPANET。“从 1972 年 6 月起，我就成了 PARC 的网络负责人。” ²³梅特卡夫将施乐的 MAXC 连接到 ARPANET，但施乐的重点是本地网络：将单用户计算机（后来成为 Alto）连接到其他类似计算机，以及连接到共享的高质量打印机，所有这些都在同一栋楼内完成。ARPANET 模型及其昂贵的专用接口消息处理器并不合适。

Ethernet was invented at Xerox-PARC in 1973 by Robert Metcalfe and David Boggs. Metcalfe was an alumnus of MIT’s Project MAC, and in 1969 he helped link MIT to ARPANET, connecting a PDP-10 to it. He moved to Xerox-PARC in 1972; one of his first tasks there was to hook up PARC’s PDP-10 clone, the MAXC, to ARPANET. “As of June 1972, I was the networking guy at PARC.”²³ Metcalfe connected Xerox’s MAXC to ARPANET, but the focus at Xerox was on local networking: to connect a single-user computer (later to become the Alto) to others like it, and to a shared, high-quality printer, all within the same building. The ARPANET model, with its expensive, dedicated Interface Message Processors was not appropriate.

梅特卡夫到达 PARC 时，已经建立了一个本地网络，使用以星型拓扑连接的 Data General 小型计算机。24^梅特卡夫和他的同事认为，即使是 Data General 网络也过于昂贵，不够灵活，无法在办公室环境中使用，因为在办公室环境中，人们可能需要频繁连接或断开机器。他还认为它不够强大——网络的运行依赖于几个关键部件不发生故障。他回忆起他在夏威夷看到的一个网络，它使用无线电信号将夏威夷群岛之间的计算机连接起来，称为 ALOHAnet。25^在这个系统中，文件被分解成不超过 1000 位长的“数据包”，每个数据包的头部都附有预期接收者的地址。网络上的其他计算机被调到 UHF 频率并监听数据包，接受发送给它的数据包并忽略所有其他数据包。

When Metcalfe arrived at PARC there was already a local network established, using Data General minicomputers linked in a star-shaped topology.²⁴ Metcalfe and his colleagues felt that even the Data General network was too expensive and not flexible enough to work in an office setting, where one may want to connect or disconnect machines frequently. He also felt it was not robust enough—the network’s operation depended on a few critical pieces not failing. He recalled a network he saw in Hawaii that used radio signals to link computers among the Hawaiian islands, called ALOHAnet.²⁵ With this system, files were broken up into “packets,” no longer than 1000 bits long, with an address of the intended recipient attached to the head of each. Other computers on the net were tuned to the UHF frequency and listened for the packets, accepting the ones that were addressed to it and ignoring all the others.

该系统之所以吸引梅特卡夫，是因为其介质（在本例中为无线电）是无源的。它只是传输信号，每个节点的计算机负责处理、排队和路由工作。施乐帕洛阿尔托研究中心的办公室之间没有水隔开，但这一概念非常适合一栋大楼内的一套办公室。梅特卡夫建议用廉价的同轴电缆代替传输 ALOHAnet 信号的“以太”。26^只需接入电缆，即可将新计算机添加到“以太网”中。要发送数据，计算机首先要进行监听，以确保线路上没有数据包；如果没有，则发送自己的数据包。如果两台计算机恰好同时传输数据，则每台计算机都会后退一个随机间隔，然后重试。如果这种冲突开始频繁发生，计算机本身就会后退，不再频繁传输数据。27^通过仔细的数学分析，梅特卡夫表明，这样的系统可以处理大量流量而不会过载。 1973 年 5 月，他撰写了一份描述，并招募了 David Boggs 来帮助构建它。到第二年，他们已经拥有了一个相当庞大的网络。梅特卡夫回忆说，它的速度大约是每秒 300 万比特，这在当时是闻所未闻的，因为“ARPANET 的 50 千比特每秒 (Kbps) 电话线路被认为很快。” ²⁸

What made this system attractive for Metcalfe was that the medium— in this case radio—was passive. It simply carried the signals, with the computers at each node doing the processing, queuing, and routing work. The offices at Xerox PARC were not separated by water, but the concept was perfectly suited for a suite of offices in a single building. Metcalfe proposed substituting a cheap coaxial cable for the “ether” that carried ALOHAnet’s signals.²⁶ A new computer could be added to the “Ethernet” simply by tapping into the cable. To send data, a computer first listened to make sure there were no packets already on the line; if not, it sent out its own. If two computers happened to transmit at the same time, each would back off for a random interval and try again. If such collisions started to occur frequently, the computers themselves would back off and not transmit so often.²⁷ By careful mathematical analysis Metcalfe showed that such a system could handle a lot of traffic without becoming overloaded. He wrote a description of it in May 1973 and recruited David Boggs to help build it. They had a fairly large network running by the following year. Metcalfe recalled that its speed, around three million bits per second, was unheard of at the time, when “the 50-kilobit-per-second (Kbps) telephone circuits of the ARPANET were considered fast.”²⁸

这些速度从根本上改变了小型计算机和大型计算机之间的关系。小型计算机集群现在终于提供了一种替代传统大型中央系统（分时并通过哑终端访问）的模式。

Those speeds fundamentally altered the relationship between small and large computers. Clusters of small computers now, finally, provided an alternative to the classic model of a large central system that was time-shared and accessed through dumb terminals.

以太网对工作站市场产生了最大影响，后来又影响到了个人电脑市场，但它的首次成功是在 1979 年，当时数字设备公司、英特尔和施乐公司联手将其确立为标准，DEC 将其用于 VAX。戈登·贝尔认为，以太网很幸运地扎根牢固，经受住了 IBM 推出的竞争方案 Token Ring 的考验。29^基于UNIX 的工作站几乎全部采用以太网，尽管也使用 Token Ring 和一些替代方案。

Ethernet would have its biggest impact on the workstation, and later PC, market, but its first success came in 1979, when Digital Equipment Corporation, Intel, and Xerox joined to establish it as a standard, with DEC using it for the VAX. Gordon Bell believes that it was fortunate in becoming rooted firmly enough to withstand the introduction of a competing scheme, Token Ring, by IBM.²⁹ UNIX-based workstations nearly all adopted Ethernet, although Token Ring and a few alternate schemes are also used.

基于 DOS 的个人计算机在实现联网方面较晚。它们使用的英特尔处理器和 DOS 都不太适合联网。还有一个社会因素在起作用：毕竟这是一台个人计算机 — — 为什么一个人要将它与别人的计算机连接起来，或者更糟的是，要共享大容量存储或打印机等资源？IBM 进入市场使个人计算机受到尊重。但许多用户并没有忘记最初催生 PC 的理想主义赋权观念。个人计算机最终也实现了联网，尽管走了一条迂回曲折的道路。

DOS-based personal computers were late in getting networking. Neither the Intel processors they used, nor DOS, was well-suited for it. There was a social factor at work, too: it was, after all, a personal computer—why should one want to connect it with someone else’s, or even worse, have to share resources like mass storage or a printer? IBM’s entry into the market made personal computers respectable. But many users had not forgotten the idealistic notions of empowerment that had spawned the PC in the first place. Personal computers eventually became networked, too, though by a circuitous route.

工作站和“VAXen”在工程师和科学家中找到了市场，但除少数例外，商业办公环境仍在使用王安、IBM 和其他公司的办公自动化系统。30^优秀的文字处理程序，尤其是 1982 年由 Lotus Development Corporation 为 IBM 个人计算机推出的电子表格程序 1-2-3，帮助 IBM PC 及其克隆产品进入该市场。Lotus 1-2-3 与 VisiCalc 类似，但有一些附加功能。（“2”和“3”表示图形和数据库功能。）由于它是专门为 IBM PC 编写的，使用汇编语言来获得更快的性能，因此它的运行速度比其他电子表格快得多。在公司办公室中，以 Apple II 为开端的个人计算机小流变成了 IBM PC 和 PC 兼容机的洪流，运行 Lotus 1-2-3、文字处理软件（如 Word Perfect）和数据库程序（如 dBase III）。

Workstations and “VAXen” found a market among engineers and scientists, but with only a few exceptions the commercial office environment continued to use office automation systems from Wang, IBM, and others.³⁰ Good word-processing programs, and especially the spreadsheet program 1-2-3, introduced for the IBM Personal Computer by Lotus Development Corporation in 1982, helped bring the IBM PC and its clones into that market. Lotus 1-2-3 was like VisiCalc, with some additional features. (The “2” and “3” implied the features of graphing and database capabilities.) Because it was written specifically for the IBM PC, using assembly language to gain faster performance, it ran much faster than other spreadsheets. In corporate offices the little stream of personal computers that began with the Apple II became a flood of IBM PCs and PC-compatibles, running Lotus 1-2-3, word processing software like Word Perfect, and database programs like dBase III.

负责公司信息服务部门的人员将这场洪水视为圣经中的瘟疫。个人电脑和个人电脑软件的购买者是出于个人需求，而非公司需求。这些个人需求往往受到行业杂志的夸张广告的驱动，这些广告承诺，任何聪明到可以购买最新软件的人，以及聪明到可以绕过公司采购和评估官僚机构的人，都可以拥有数字乌托邦。信息服务人员沉迷于王安字处理器或 IBM 大型机，失去了对他们受雇管理的内容的控制。

The people running corporate information services departments saw this flood as a Biblical plague. Purchasers of PCs and PC software were driven by personal, not corporate, needs. These personal needs were often driven by advertising hyperbole in the trade journals, which promised digital Utopia for anyone smart enough to buy the latest piece of software, and smart enough to bypass the bureaucracy of corporate purchasing and evaluation. Information services people, wedded to Wang word processors or IBM mainframes, were losing control of what they were hired to manage.

双方都说得对。PC 和 DOS 标准催生出的商业软件不仅价格低廉，而且比集中式系统更好。PC 也催生出质量低劣的软件，浪费的公司时间比节省的时间还多。PC 中的“个人”一词还意味着，员工基于个人满意度对软件的选择并不总是与其雇主的目标相符。

Both sides were right. The PC and DOS standards led to commercial software that was not only inexpensive but also better than what came with centralized systems. The PC also led to poor quality software that wasted more company time than it saved. The “personal” in the PC meant also that a worker’s choices of software, based on personal satisfaction, did not always mesh with the goals of his or her employer.

到 20 世纪 80 年代中期，很明显，无论多少公司政策指令都无法阻止 PC 进入办公室，尤其是对那些已经在家中拥有 PC 的员工而言。解决方案是技术修复：通过局域网 (LAN) 将 PC 相互联网。到 1984 年，已有超过 20 种产品发布或上市，声称可以通过局域网连接 IBM PC。31^由于早期 IBM 个人电脑和克隆机所采用的英特尔处理器的限制，其中许多产品性能不佳。在更先进的英特尔芯片版本普及之后（尤其是 1985 年推出的 80386），市场经历了一场洗牌，PC 联网变得切实可行。到 1989 年，占据了一半以上业务的公司是位于盐湖城地区的 Novell。 Novell 的 Netware 是一个复杂且昂贵的操作系统，它覆盖在 DOS 之上，控制着机器并将控制权交给“文件服务器”——通常是具有大量存储和 I=O 能力的 PC（“服务器”一词源于 Metcalfe 和 Boggs 1976 年关于以太网的论文）。通过将数据和办公自动化软件放在此服务器上而不是单独的机器上，可以重新建立某种程度的中央控制。

By the mid-1980s it was clear that no amount of corporate policy directives could keep the PC out of the office, especially among those employees who already had a PC at home. The solution was a technical fix: network the PCs to one another, in a local-area network (LAN). By 1984 there were over twenty products announced or on the market that claimed to connect IBM PCs in a local-area network.³¹ Many of these performed poorly, given the limits of the Intel processor that powered early IBM Personal Computers and clones. After more advanced versions of the Intel chip became common (especially the 80386, introduced in 1985), there was a shakeout, and networking for PCs became practical. The company that emerged with over half the business by 1989 was Novell, located in the Salt Lake City area. Novell’s Netware was a complex—and expensive—operating system that overlaid DOS, seizing the machine and directing control to a “file server”—typically a PC with generous mass storage and I=O capability (the term “server” originated in Metcalfe and Boggs’s 1976 paper on Ethernet). By locating data and office automation software on this server rather than on individual machines, some measure of central control could be reestablished.

从一开始，个人电脑的联网就落后于 UNIX 工作站，但个人电脑的低成本和更好的办公软件推动了这一市场的发展。有人预测 PC 和 UNIX 工作站将融合，但这并没有发生。办公室工作人员使用电脑进行文字处理，使用电子表格进行基本会计工作，而那些归档数据的人则使用简单的数据库程序。对于 PC，有大量适用于这些应用程序的优质软件可供选择，价格合理。那些需要图形（例如桌面出版）的人可以购买 Macintosh。人们没有动力花更多的钱购买 UNIX 工作站，因为它提供的日常软件较少。回想一下，UNIX 更像是一套工具，而不是一个完整的操作系统。它甚至比 DOS 更难使用和理解。UNIX 也没有那么标准化。包括 X-Windows 在内的几种窗口系统于 20 世纪 80 年代末进入 UNIX 世界，但没有出现与 Microsoft Windows 相比的 DOS 机器标准^。32

Networking of PCs lagged behind the networking that UNIX workstations enjoyed from the start, but the personal computer’s lower cost and better office software drove this market. Some predicted a convergence of PC and UNIX workstations, but that did not occur. Office workers used computers for word processing, basic accounting using a spreadsheet, and those who filed data used simple database programs. There was an abundant selection of good software for these applications, at reasonable prices, for the PC. Those who needed graphics, say for desktop publishing, could buy a Macintosh. There was no incentive to spend more money for a UNIX workstation, which offered less workaday software. Recall that UNIX was more a set of tools than a complete operating system. And it was even harder to use and understand than DOS. Nor was UNIX as standardized. Several windowing systems, including X-Windows, came to the UNIX world in the late 1980s, but no standard emerged that compared to Microsoft Windows for DOS machines.³²

尽管 IBM PC 架构和 MS-DOS 饱受诟病，但这些标准能够发展到今天，不仅可行，而且长期占据主导地位，这本身就是一种质量的体现。到 1995 年，该标准仍然包括英特尔 8086 处理器的后代、由 Microsoft Windows 3.1 版或更高版本增强的 DOS 高级版本，以及 Novell 或少数其他公司的网络方案。Macintosh 也不断发展并保持竞争力，尽管面对 IBM 兼容机的高级 Windows 版本的竞争，其市场份额有所下降。

For all the criticisms of the IBM PC architecture and of MS-DOS, it is a measure of their quality that those standards could evolve so far and remain not just viable but dominant for so long. By 1995 the standard still included a descendant of the Intel 8086 processor, an advanced version of DOS enhanced by Microsoft Windows, version 3.1 or later, and a networking scheme from Novell or a handful of other companies. The Macintosh also evolved and remained competitive, although its share of the market declined in the face of competition from advanced versions of Windows for the IBM-compatible machines.

局域网络至少在办公室环境中将“个人”从个人计算中剔除。（人们仍然可以在家里做任何想做的事情。）工作场所的 PC 用户接受了这种浮士德式的交易。他们中更精通计算机的人会抵制，但大多数办公室工作人员几乎没有注意到，这代表着与最初推动个人计算机发明的力量的转变。这种转变发生得如此轻松，表明那些相信真正自主的个人计算的人可能太天真了。尽管如此，20 世纪 90 年代联网的办公室计算机为用户提供了比 20 世纪 70 年代通过“哑终端”或“玻璃电传打字机”访问的分时主机更多的自主权和独立性。这与《Byte》杂志或 Homebrew 计算机俱乐部的人们想象中的情况完全不同。

Local networking took the “personal” out of personal computing, at least in the office environment. (One could still do whatever one wanted at home.) PC users in the workplace accepted this Faustian bargain. The more computer-savvy among them resisted, but the majority of office workers hardly even noticed how much this represented a shift away from the forces that drove the invention of the personal computer in the first place. The ease with which this transition took place shows that those who believed in truly autonomous, personal computing were perhaps naive. Still, the networked office computers of the 1990s gave their users a lot more autonomy and independence than the time-shared mainframes accessed through “dumb terminals” or “glass Teletypes” in the 1970s. It was just not how the people at Byte magazine or the Homebrew Computer Club had imagined things would evolve.

网络 II：互联网

Networking II: Internet

将办公室工作人员连接到 LAN 的大部分好处都归于管理员和经理。对于他们而言，用户不再需要担心备份文件（反正很少有 PC 用户学会忠实地备份文件），而且他们现在可以使用电子邮件相互交换文件和消息。但是，连接到 LAN 的用户还获得了一个意想不到的、非常重要的回报——访问互联网。

Most benefits of connecting office workers to a LAN went to administrators and managers. For their part, users no longer had to worry about backing up files—something few PC owners ever learned to do faithfully anyway—and they could now exchange files and messages with one another using electronic mail. But there was one unanticipated, very important thing that users connected to a LAN got in return—access to the Internet.

当今的互联网虽然广为人知，但很难定义。它起源于第 6 章中描述的 ARPANET （图 9.3）。与 ARPANET 和前面描述的其他网络一样，互联网使用“分组交换”。发送消息不需要从一台计算机到另一台计算机的专用连接，就像打电话一样。33 但是，互联网^有几个主要区别。互联网不是一个单一的网络，而是全球许多不同网络的连接；因此得名。其中一些网络向公众开放，而不仅仅是向受限制或特权的社区开放。（请注意，仍有许多网络受到限制，例如银行用于内部运营的网络。）最后，互联网通过使用通用协议 TCP=IP（传输控制协议 = 互联网协议）允许跨这些不同网络进行通信。这种使用 TCP=IP 粘合剂的网络互连构成了当今的互联网^。34

The present-day Internet, though well known, is hard to define. It is descended from the ARPANET described in chapter 6 (figure 9.3). Like ARPANET and the other networks described earlier, the Internet uses “packet switching.” Sending a message does not require a dedicated connection from one computer to another, as, say, one has when calling someone on the telephone.³³ There are however several major differences. The Internet is not a single network but rather the connection of many different networks across the globe; hence the name. Some of those networks are open to the public, not just to a restricted or privileged community. (Note there are still many networks that are restricted, e.g., one used by a bank for its internal operations.) Finally, the Internet allows communication across these different networks by its use of a common protocol, TCP=IP (transmission control protocol= internet protocol). This interconnection of networks to one another, using the glue of TCP=IP, constitutes the present-day Internet.³⁴

互联网之所以能得到广泛使用，是社会和技术因素的共同作用的结果。前者包括 20 世纪 80 年代，资金和行政支持从 ARPA 转移到国家科学基金会，然后在 20 世纪 90 年代转移到允许任何人（包括那些将其用于商业目的的人）访问互联网的实体。直到 1992 年，互联网用户还大致均匀地分布在政府、教育、军事、网络相关、商业和非营利组织之间（其地址分别由“.gov”、“.edu”、“.mil”、“.net”、“.com”和“.org”序列标识）。到 1995 年，商业用户压倒了其他用户，短语“'X' dot com”（其中“X”是一家公司的名称）已进入我们的词汇表。记录这一事件序列的工作仍在进行中，互联网发展的这一方面将不再进一步讨论^。35

The Internet made its way into general use by a combination of social and technical factors. Among the former was the shift of financial and administrative support from ARPA, to the National Science Foundation in the 1980s, and then in the 1990s to entities that allowed Internet access to anyone, including those who would use it for commercial purposes. As recently as 1992, Internet users were about evenly distributed among governmental, educational, military, net-related, commercial, and nonprofit organizations (identified in their addresses, respectively, by the sequences “.gov,” “.edu,” “.mil,” “.net,” “.com,” and “.org”). By 1995 commercial users overwhelmed the rest, and the phrase “‘X’ dot com,” where “X” is the name of a corporation, has entered our vocabulary. Documenting this sequence of events is a work still in progress and that aspect of the Internet’s growth will not be discussed further here.³⁵

图 9.3

ARPANET，截至 1970 年 12 月。地图显示 11 个节点，主要集中在洛杉矶和波士顿地区。（来源：DARPA。）

Figure 9.3

ARPANET, as of December 1970. The map shows 11 nodes, mainly concentrated in the Los Angeles and Boston regions. (Source : DARPA.)

互联网出现背后的技术因素更为人所知。其中一个因素已经提到：ARPA 支持 TCP=IP 协议的开发，并于 1980 年决定采用该协议。ARPA 的支持以及该协议被纳入伯克利 UNIX 意味着它将在大学中广泛使用，并且不会被 IBM、DEC 或任何其他计算机公司视为专有产品^。36

The technical factors behind the emergence of the Internet are better known. One has already been mentioned: ARPA’s support for the development of, and its decision in 1980 to adopt, the TCP=IP protocol. ARPA’s support, and the protocol’s inclusion in Berkeley UNIX, meant that it would become widely available in universities and would not be held as a proprietary product by IBM, DEC, or any other computer company.³⁶

另一个因素是局域网的兴起。ARPANET 的最初目标是共享昂贵的计算机资源；个人可以通过分时连接到大型机的终端访问这些资源。（随着 ARPANET 的形成，人们还可以在一些地方通过终端接口处理器 (TIP) 将终端直接连接到网络 - 这是 IMP 概念的变体。）随着 1973 年以太网的发明以及次年个人计算机的发明，计算的经济性发生了变化。计算能力不再稀缺。分时技术逐渐成熟并可用于许多大型机，但它被 Xerox-PARC 工作中衍生的客户端-服务器计算所取代。而在整个这个时代，摩尔定律占据了主导地位：计算能力（以这些机器中硅片的密度来衡量）大约每 18 个月翻一番。

Another factor was the rise of local area networks. The initial goal for ARPANET was the sharing of expensive computer resources; individuals would gain access to these resources through terminals that were connected to mainframes by time-sharing. (As the ARPANET took shape one could also, in a few places, connect a terminal directly to the network through a terminal interface processor (TIP)—a variation of the IMP concept.) With the invention of Ethernet in 1973, and the personal computer the following year, the economics of computing changed. Computing power was no longer scarce. Time-sharing matured and became available on many mainframes, but it was supplanted by client-server computing that descended from the work at Xerox-PARC. And throughout this era Moore’s Law ruled: computing power, as measured by the density of the silicon chips that went into these machines, was doubling about every eighteen months.

局域网使大量人能够访问互联网。以太网的速度足以与构成互联网主干的专用线路的高速度相媲美。高速网络一直是工作站公司想要提供的功能之一——回想一下 SUN 的营销口号：“网络就是计算机。”人们没有预料到的是，在英特尔越来越强大的处理器的推动下，个人计算机的发展将这种能力带到了学术和研究领域以外的办公室和其他地方。到 20 世纪 80 年代末，拥有 UNIX 工作站的人，到 1995 年，拥有 LAN 上个人计算机的人，都可以访问互联网，而每台机器都不需要直接连接到互联网的高速线路。

Local area networks made it possible for large numbers of people to gain access to the Internet. Ethernet’s speeds were fast enough to match the high speeds of the dedicated lines that formed the Internet’s backbone. High-speed networking had always been among the features workstation companies wanted to supply—recall SUN’s marketing slogan: “The Network is the Computer.” What had not been anticipated was how advances in personal computers, driven by ever more powerful processors from Intel, brought that capability to offices and other places outside the academic and research worlds. By the late 1980s those with UNIX workstations, and by 1995 those with personal computers on a LAN, all had access to the Internet, without each machine requiring a direct connection to the Internet’s high-speed lines.

因此，以太网的高数据速率提供了一种解决方法，即通信速度和数据容量未能跟上计算机处理速度和存储速度的进步。戈登·摩尔在英特尔的同事安德鲁·格罗夫提出了自己的“定律”来描述这种差距：芯片密度每 18 个月翻一番（摩尔定律），而电信带宽每 100 年翻一番（格罗夫定律）。37^自1990 年以来，带宽增长有所加快，但仍然远远落后于芯片密度的增长。格罗夫认为，这种差距是电信行业过度监管的结果。无论原因是什么，进入家庭和办公室的电话线无法以每秒超过 5 万比特的速度处理数据——通常更低，无论如何都远低于以太网实现的速度和许多互联网活动所需的速度。

Ethernet’s high data rates thus provided a way of getting around the fact that communication speeds and data capacity had not kept up with the advances in computer processing speeds and storage. Gordon Moore’s colleague at Intel, Andrew Grove, came up with his own “law” to describe this disparity: while chip density doubles every eighteen months (Moore’s Law), telecommunications bandwidth doubles every 100 years (Grove’s Law).³⁷ Bandwidth growth has picked up since 1990, but it still lags well behind the growth of chip density. Grove believes the disparity is a result of overregulation of the telecommunications industry. Whatever the cause, it is true that the telephone lines coming into homes and offices cannot handle data at more than about 50 thousand bits per second—usually less, and in any case well below the speeds achieved by Ethernet and required by many Internet activities.

自 20 世纪 90 年代中期以来，调制解调器制造商为提高普通电话连接的数据速率做出了巨大努力。也有计划使用家庭中的其他线路进行互联网通信：有线电视线路甚至电力线。各种卫星或微波无线技术也正在开发中。其中一种方法可能会打破瓶颈。与此同时，人们确实通过拨打本地电话号码将家用电脑连接到互联网，但他们获得的接入量只是办公室、实验室或大学所能获得的接入量的一小部分。目前，格罗夫定律似乎仍然成立。

Since the mid-1990s modem manufacturers have made heroic efforts to improve data rates for ordinary telephone connections. Plans have also emerged to use other wires that come into the home for Internet traffic: the line that carries cable television or even the power lines. Various satellite or microwave wireless technologies are also being developed. One of these approaches will probably break the bottleneck. Meanwhile, people do connect their home computers to the Internet by dialing a local telephone number, but the access they get is a fraction of what they can find at the office, laboratory, or university. For now, Grove’s Law seems to hold.

网络 III：万维网

Networking III: the World Wide Web

随着互联网从 ARPA 中诞生，它开始发生变化。互联网上的最初活动源自 ARPANET：用户可以登录远程计算机，将大型文件从一台机器传输到另一台机器，并发送邮件。最初 ARPANET 的明确目标是后来被称为“Telnet”和“FTP”的前两项活动；邮件不是，但它在最初几个节点工作后不久就出现了。早期的电子邮件功能被嫁接到文件传输操作上，但不久之后就开发出了专用的电子邮件软件，其中已经具备了现代电子邮件系统中的大多数功能（例如，回复某人、保存消息或向列表发送消息的能力）^。38

As the Internet emerged from its roots in ARPA, it began to change. The initial activities on the Internet were ARPANET derived: users could log on to a remote computer, transfer large files from one machine to another, and send mail. The first two activities later known as (“Telnet” and “FTP”) were explicit goals of the original ARPANET; mail was not, but it emerged soon after the first few nodes were working. Early e-mail facilities were grafted onto the file-transfer operation, but before long dedicated e-mail software was developed, with most of the features found in modern e-mail systems already in place (e.g. the ability to reply to someone, save a message, or send a message to a list).³⁸

对该三元组的第一个重要扩展暗示了大众媒体所称的基于互联网的“虚拟社区”。这个短语是否有意义，如果有，它是什么，将在后面讨论，但引发它出现的是早期互联网上新闻或讨论组的发展。尽管这些群组与互联网有关，但多年来只有那些能够访问 UNIX 系统的人才能访问它们。对于普通大众来说，它们在个人计算机领域是由所谓的公告板系统 (BBS) 所引领的，顾名思义，它就像公告板一样，任何人都可以在上面发布供所有人阅读的公告。BBS 通常在有限的设施上运行，使用在 IBM XT 或同等产品上运行的公共域软件。关键的技术发展是 Hayes Microcomputer Products 于 1981 年推出的廉价调制解调器，以及 1983 年推出的带有 10 兆字节硬盘的 XT 本身。用户通过本地电话拨号进入这些公告板，速率为每秒几百比特。³⁹但它们却发挥了作用，而且很受欢迎，有些甚至一直使用到 20 世纪 90 年代。

The first serious extension to that triad gave a hint of what the popular press calls a “virtual community” based on the Internet. Whether that phrase has any meaning, and if so, what it is, will be discussed later, but what evoked it was the development of news or discussion groups on the early Internet. Although these groups are associated with the Internet, for years only those with access to UNIX systems had access to them. For the general public they were anticipated in the personal computer arena by so-called bulletin-board systems (BBSs), which as the name implies, acted like bulletin boards, on which anyone could post a note for all to read. BBSs typically ran on limited facilities, using public-domain software running on an IBM XT or the equivalent. Key technical developments were the introduction in 1981 of an inexpensive modem by Hayes Microcomputer Products, and of the XT itself in 1983, with its 10megabyte hard disk. Users dialed into these bulletin boards with a local telephone call, at a rate of a few hundred bits per second.³⁹ But they worked and were well liked, and some remained in use into the 1990s.

基于 UNIX 的新闻组最早出现在 1979 年之后，与主流 ARPANET-Internet 活动有些独立，通用名称为 Usenet。40^这些新闻组被分为一组主要类别，例如“comp”代表计算机，“rec”代表爱好。一个名为“alt”（另类）的类别包括倾向于怪异的子类别；当然，这些是报纸记者在印刷的第一批有关互联网的报道中关注的。记者知道，关于另类性取向的报道会比关于 UNIX 文件结构的报道吸引更多的读者，即使后者对网络的运行更为重要。也许编辑们有点担心，如果不加以控制，新闻组会让他们破产（他们可能会）。由此产生的故事几乎不准确，但确实激发了人们对互联网的普遍兴趣，尽管当时很少有大学以外的人能够访问这些组。

UNIX-based news groups first appeared after 1979, somewhat independently of the mainstream ARPANET-Internet activities and under the general name of Usenet.⁴⁰ These were arranged into a set of major categories, for example, “comp” for computers, or “rec” for hobbies. A category called “alt” (“alternative”) included subcategories that tended to the bizarre; of course these were the ones that newspaper reporters focused on in the first stories about the Internet to appear in print. Reporters knew that a story about alternative sexual preferences would attract more readers than a story about UNIX file structures, even if the latter were far more important to the operation of the Net. And perhaps editors were somewhat fearful that, if unchecked, news groups would put them out of business (they might). The resulting stories were hardly accurate but did kindle a general interest in the Internet, although at a time when few people outside universities could gain access to these groups.

另一种公告板系统出现在使用 IBM 大型机的大学和工业计算机中心。大约从 1981 年开始，纽约城市大学和耶鲁大学的 IBM 系统管理员开发了 BITNET，它通过巧妙的编程将大型 IBM 系统连接起来，将消息视为打在 80 列卡片上的数据。BITNET 带来了 LISTSERV，这是一种管理讨论组的系统。正如 IBM 大型机用户所预料的那样，LISTSERV 讨论比 Usenet 上的讨论更正式、更严肃，尽管它们的范围也很广。经过大约十五年的并行运营，这三个流融合成一个新闻、讨论、实时聊天和其他信息交换场所的社区，现在可在万维网上找到。

Another bulletin board system emerged at university and industrial computer centers that used IBM mainframes. Beginning around 1981, IBM system administrators at the City University of New York and Yale developed BITNET, which linked large IBM systems by a clever bit of programming that treated messages as though they were data punched on decks of 80-column cards. With BITNET came LISTSERV, a system for managing discussion groups. As might be expected among IBM mainframe users, LISTSERV discussions were a little more formal and serious than those on Usenet, although they also had a wide range. After about a decade and a half of parallel operations, all three of these streams blended into a community of news, discussions, real-time chat, and other venues for exchanging information, now found on the World Wide Web.

地鼠，WAIS

Gopher, WAIS

如果用户已经知道所需文件的位置和名称，则文件传输和远程登录没有问题。也许该信息可以发布在新闻组上以响应查询，但随着互联网的发展，互联网上可用的信息量使这些设施不堪重负。互联网开始感觉像一个没有卡片目录的大型图书馆，或者一个二手书店，里面有大量便宜的好书，但书却随意堆放在书架上。 1990 年或 1991 年初，明尼苏达大学的程序员为此创建了 Gopher，以纪念大学吉祥物。Gopher 最初允许学生和教职员工（包括那些使用计算机经验不多的人）查询校园计算机以获取课程表、管理政策声明和体育赛事等信息，然后 Gopher 会“获取”数据并将其传递给坐在终端前的人。它很快传播到其他地方，系统管理员在连接到互联网的机器上安装了该程序。如果 Gopher 安装在本地站点，用户可以调用该程序并直接使用它；如果没有，他们可以通过 telnet 连接到一个拥有 Gopher 的远程站点，并允许外部人员使用它（很快就出现了几个地方）。该程序以菜单的形式显示信息，采用分层树结构。最终，用户可以获取所需的数据（例如，本地天气预报，从树的顶部开始，例如“国家气象局”）。使用 Gopher 很繁琐，但它确实有效，并很快被接受。

File transfer and remote log-in were fine if users already knew where a desired file was located and what it was called. Perhaps that information might be posted on a news group in response to a query, but as the Internet grew, the amount of information available on it overwhelmed these facilities. The Internet began to feel like a large library that had no card catalog, or a used bookstore that had an incredible number of great books at bargain prices but with the books piled at random on the shelves. In 1990 or early 1991, programmers at the University of Minnesota responded by creating Gopher, in honor of the university mascot. Gopher at first allowed students and faculty, including those with little experience using computers, to query campus computers for information such as class schedules, administrative policy statements, and sporting events, Gopher would then “go fer” the data and deliver it to the person seated at the terminal. It soon spread elsewhere, where system administrators installed the program on machines connected to the Internet. If Gopher was installed at a local site, users could call the program and use it directly; if not, they could telnet to a distant site that had Gopher and allowed outsiders to use it (several places soon emerged). The program displayed information as menus of menus, in a hierarchical tree. Eventually users could get to the desired data (e.g., the local weather forecast, starting at the top of the tree with, say, “National Weather Service”). Using Gopher was tedious, but it did work and was quickly accepted.

Gopher 的菜单列表有时会陷入死胡同，很难再回头。这让不止一个人“迷失在网络空间”。Gopher 列出的文件名可能不是用户想要的名称（例如，“当地天气”，在 Gopher 上列出的是“区域雷达图像”）。在剑桥超级计算机公司 Thinking Machines，Brewster Kahle 和他的同事开发了一个名为 WAIS（广域信息服务）的系统，允许用户直接搜索文件内容。WAIS 搜索的文档中每个单词都构建了索引。这乍一看可能令人望而生畏，但到 1990 年，已经有几种可用于构建此类索引的 PC 文字处理程序，而 Thinking Machines 销售的先进计算机旨在执行这种重型处理。WAIS 的学习和使用并不比 Gopher 难多少，但它从未像 Gopher 那样普及。当它正常工作并检索到用户想要查找的内容时，这位幸运的用户立即意识到这就是互联网的意义所在。

Gopher’s lists of menus sometimes led to a dead end, from which it was hard to backtrack. That left more than one person “lost in cyberspace.” The file name that Gopher listed might not be the name a user had in mind (e.g., “local weather,” listed on Gopher as “regional radar image”). At the Cambridge supercomputer company Thinking Machines, Brewster Kahle and his colleagues developed a system called WAIS (Wide Area Information Service) that allowed users to search the contents of files directly. WAIS searched documents for which an index of every word was constructed. This may have at first seemed absurdly daunting, but by 1990 there were already several word-processing programs available for PCs that constructed such indexes, and the advanced computers sold by Thinking Machines were intended to do just that kind of heavy-duty processing. WAIS was not much harder to learn or to use than Gopher, but it never become as widespread. When it worked and retrieved exactly what a user wanted to find, the fortunate user got an immediate sense that this was what the Internet was all about.

和早期的新闻组一样，Gopher 和 WAIS 也被万维网及其信息检索系统淘汰了。它们在 1995 年仍然可以找到，但那时，就像 Telnet、FTP 和 LISTSERV 一样，这些索引程序已不再是使用互联网的核心。1995 年后使用互联网的大多数人可能甚至都没有听说过它们。人们可以很容易地在不提及 Gopher 的情况下撰写互联网的历史，因为它出现的时间太短了。但在本章中，它被提及了，尽管只是简短的提及，原因与前面几章（更详细地）讨论打孔卡的原因相同。也许有一天，万维网也会被遗忘，因为它也被新的东西取代了。

Like the early news groups, Gopher and WAIS were rendered obsolete by the World Wide Web and its system of information retrieval. They could still be found in 1995, but by then, as with Telnet, FTP, and LISTSERV, these indexing programs were no longer central to using the Internet. Most people who used the Net after 1995 had probably not even heard of them. One could easily write a history of the Internet without mentioning Gopher at all, so brief was its time on the stage. But it is mentioned, if only briefly, in this chapter, for the same reason that earlier chapters dealt (at greater length) with punched cards. Perhaps someday the World Wide Web will be forgotten after it, too, has been replaced by something new.

万维网，马赛克

World Wide Web, Mosaic

讨论万维网时，很难不去面对这样一种普遍的感觉：万维网是自 1945 年冯·诺依曼的 EDVAC 报告以来计算领域所有发展的顶峰。或者，它是自活字印刷术发明以来通信领域进步的顶峰？还是书写的顶峰？随你选择吧。描述万维网的历史（其初始代码是在 1990 年编写的）要比描述计算的早期历史困难得多。历史学家避免描述最近的过去是有充分理由的。如果历史以“所有计算历史的顶峰是 Gopher”这一说法结束，读者会怎么看待这段历史？

It is difficult to discuss the World Wide Web without confronting a general feeling that this is the culmination of all the developments in computing since von Neumann’s EDVAC Report of 1945. Or is it the culmination of advances in communication since the invention of printing with movable type? Or the invention of writing? Take your pick. Giving a history of the World Wide Web, the initial code for which was written in 1990, will be much more difficult than writing about the early days of computing. Historians avoid writing about the recent past for good reasons. What would readers think of a history that ended with a statement that the culmination of all computing history was Gopher?

与此同时，网络之所以吸引历史学家，是因为它根深蒂固。它之所以吸引人，还因为它阐明了这一叙述的中心主题，即计算有时是如何通过几乎随机的事件而取得进展的。我们已经看到，计算有时是如何通过像埃克特和莫奇利这样的发明家的远见、精力和决心而向前发展的，他们几乎将未来变成了现实。在其他时候，它的进展就像是偶然的，例如，阿尔伯克基的一家模型火箭爱好者商店推出了 Altair。

At the same time, the Web attracts the historian because its roots are so deep. It is attractive also for the way it illustrates a central theme of this narrative, of how computing progresses at times by almost random events. We have seen how computing was moved along at times by the vision, energy, and determination of inventors like Eckert and Mauchly, who almost willed the future into being. At other times it has proceeded as if by accident, for example, the introduction of the Altair by a model-rocket hobby shop in Albuquerque.

万维网的发展既有随机性，也有计划性。它的发明地点完全出乎意料：位于瑞士和法国边境的高能物理实验室 CERN。它不是来自 IBM、施乐或微软的研究实验室，也不是来自著名的麻省理工学院媒体实验室，后者曾大张旗鼓地宣称其主要目标是通过整合计算机和通信来改变世界。然而，网络的基本概念，即将信息构建为“超文本”，可以追溯到 1945 年万尼瓦尔·布什 (Vannevar Bush) 的一篇开创性论文，该论文讨论了即将到来的信息过剩以及如何应用技术来处理它。41 布什的论文影响了后来几代计算机研究人员。其中两个人在前几章中提到过：道格·恩格尔巴特 (Doug Engelbart)，他在 1968 年秋季联合计算机会议上展示了一个原型信息检索系统，以及泰德^·尼尔森 (Ted Nelson)，他是《Computer Lib = Dream Machines》的作者，也是类似系统 Xanadu 的开发者。^四十二

The development of the World Wide Web has elements of both randomness and planning. It was invented at an unforeseen and totally unexpected place: the high-energy physics laboratory CERN, on the Swiss-French border. It did not come from the research laboratories of IBM, Xerox, or even Microsoft, nor did it come out of the famed Media Lab at MIT, which stated with great fanfare that its primary goal was to transform the world through the integration of computers and communications. Yet the Web’s fundamental concept, of structuring information as “hypertext,” goes back to a seminal essay by Vannevar Bush in 1945, about the coming glut of information and how technology might be applied to handle it.⁴¹ Bush’s essay influenced subsequent generations of computer researchers. Two of those people have been mentioned in previous chapters: Doug Engelbart, who demonstrated a prototype information retrieval system at the 1968 Fall Joint Computer Conference, and Ted Nelson, the author of Computer Lib=Dream Machines and developer of a similar system called Xanadu.⁴²

在他自出版的宣言中，尼尔森将“超文本”定义为“根据要求分支或执行的写作形式；它们最适合在计算机显示屏上呈现。” ⁴³^尼尔森称赞了恩格尔巴特的在线系统（NLS），但指出恩格尔巴特相信以大纲格式紧密构建信息。44尼尔森想要更接近万尼瓦尔·布什早期概念的东西，布什希望这个概念能够复制大脑跨学科界限进行联想的能力。在 20 世纪 70 年代和 80 年代，尼尔森不知疲倦地工作，让“上都”（Xanadu）成为现实。他与学术和研究界保持着密切联系，但始终置身于学术和研究界之外，他的想法启发了布朗大学安德里斯·范达姆 (Andries van Dam) 领导的工作。⁴⁵苹果公司独立于这些研究人员于 1987 年为 Macintosh 推出了一款名为 HyperCard 的程序。Hyper-Card 仅实现了范达姆或尼尔森理解的超文本概念的一小部分，但它简单易用，甚至初学者也可以轻松编程。尽管存在种种限制，但 HyperCard 将非线性文本和图形的概念带出了实验室环境。

In his self-published manifesto, Nelson defined “hypertext” as “forms of writing which branch or perform on request; they are best presented on computer display screens.”⁴³ Nelson praised Engelbart’s On-Line System (NLS) but noted that Engelbart believed in tightly structuring information in outline formats.⁴⁴ Nelson wanted something closer to Vannevar Bush’s earlier concept, which Bush hoped would replicate the mind’s ability to make associations across subject boundaries. Nelson worked tirelessly through the 1970s and 1980s to bring Xanadu to life. He remained close to, but always outside of the academic and research community, and his ideas inspired work at Brown University, led by Andries van Dam.⁴⁵ Independently of these researchers, Apple introduced a program called HyperCard for the Macintosh in 1987. Hyper-Card implemented only a fraction of the concepts of hypertext as van Dam or Nelson understood the concept, but it was simple, easy to use, and even easy for a novice to program. For all its limits, HyperCard brought the notion of nonlinear text and graphics out of the laboratory setting.

在这其中，互联网应运而生，人们突然需要找到一种方法来浏览其丰富且不断增长的资源。46^现在写下接下来发生的事情还为时过早。蒂姆·伯纳斯-李 (Tim Berners-Lee) 于 1990 年底编写了最初的万维网原型，他曾写过一本简短的回忆录来记录那段时光，但完整的故事尚未讲述。47^伯纳斯-李在欧洲粒子物理实验室 CERN 工作期间开发了万维网。他表示：“万维网的主要目标是成为一个共享的信息空间，人与机器可以通过它进行通信。这个空间应该是包容的，而不是排他性的。” ⁴⁸他特别关注如何实现不同类型计算机和软件之间的通信。他还想避免大多数数据库的结构，这种结构迫使人们在知道这种分类是否合适之前就将信息分类。为此，他设计了一种通用资源标识符（后来称为统一资源定位符或 URL），可以“指向信息世界中的任何文档（或任何其他类型的资源）。” ⁴⁹他创建了一种更复杂的超文本传输协议 (HTTP)，以取代当时使用的文件传输协议，该协议速度更快，功能更多。最后，他为超文本在网络上的移动定义了一种超文本标记语言 (HTML)。几年之内，这些缩写词以及万维网本身的 WWW 就变得像 RAM、K 或计算机领域的任何其他术语一样普遍。

In the midst of all that sprouted the Internet, with a sudden and unexpected need for a way to navigate through its rich and ever-increasing resources.⁴⁶ It is still too early to write the history of what happened next. Tim Berners-Lee, who wrote the original Web prototype in late 1990, has written a brief memoir of that time, but the full story has yet to be told.⁴⁷ Berners-Lee developed the Web while at CERN, the European particle physics laboratory. He stated that “[t]he Web’s major goal was to be a shared information space through which people and machines could communicate. This space was to be inclusive, rather than exclusive.”⁴⁸ He was especially concerned with allowing communication across computers and software of different types. He also wanted to avoid the structure of most databases, which forced people to put information into categories before they knew if such classifications were appropriate or not. To these ends he devised a Universal Resource Identifier (later called the Uniform Resource Locator or URL) that could “point to any document (or any other type of resource) in the universe of information.”⁴⁹ In place of the File Transfer Protocol then in use, he created a more sophisticated Hypertext Transfer Protocol (HTTP), which was faster and had more features. Finally, he defined a Hypertext Markup Language (HTML) for the movement of hypertext across the network. Within a few years, these abbreviations, along with WWW for the World Wide Web itself, would be as common as RAM, K, or any other jargon in the computer field.

万维网起步缓慢。它的独特功能，即通过超链接跳转到不同资源的能力，在除 CERN 之外至少有其他几个地方支持它之前，它几乎没有用处。在编写编辑软件之前，用户必须手动构建文档中的链接，这是一个非常繁琐的过程。要查看 Web 资料，人们使用一种称为“浏览器”的程序（该术语可能源自 Apple 的 Hypercard）。早期的 Web 浏览器（包括两款名为 Lynx 和 Viola 的浏览器）显示的屏幕与 Gopher 的类似，带有菜单选择列表。

The World Wide Web got off to a slow start. Its distinctive feature, the ability to jump to different resources through hyperlinks, was of little use until there were at least a few other places besides CERN that supported it. Until editing software was written, users had to construct the links in a document by hand, a very tedious process. To view Web materials one used a program called a “browser” (the term may have originated with Apple’s Hypercard). Early Web browsers (including two called Lynx and Viola) presented screens that were similar to Gopher’s, with a lists of menu selections.

1992 年秋，马克·安德森和埃里克·比纳开始讨论如何让浏览网络更加容易。安德森还在伊利诺伊大学读书时，就为美国国家超级计算应用中心做编程工作，该中心由美国国家科学基金会资助，在校园内设立，旨在让超级计算更加普及（参见最初的 ARPANET 的推动力）。1993 年 1 月，安德森和比纳编写了浏览器的早期版本，后来他们将其称为 Mosaic，并在互联网上发布了该浏览器的一个版本。50 ^Mosaic将 Hypercard 的易用性与万维网的完整超文本功能结合在一起。要选择项目，人们可以使用鼠标（因此回到了为此目的而发明该浏览器的道格·恩格尔巴特）。人们可以通过不同颜色知道某个项目有超链接。Mosaic 的第二个功能，也是最让首次使用它的人印象深刻的功能，是它将文本和图像无缝集成。

Around the fall of 1992 Marc Andreessen and Eric Bina began discussing ways of making it easier to navigate the Web. While still a student at the University of Illinois, Andreessen took a job programming for the National Center for Supercomputing Applications, a center set up with NSF money on the campus to make supercomputing more accessible (cf. the impetus for the original ARPANET). By January 1993 Andreessen and Bina had written an early version of a browser they would later call Mosaic, and they released a version of it over the Internet.⁵⁰ Mosaic married the ease of use of Hypercard with the full hypertext capabilities of the World Wide Web. To select items one used a mouse (thus circling back to Doug Engelbart, who invented it for that purpose). One knew an item had a hyperlink by its different color. A second feature of Mosaic, the one that most impressed the people who first used it, was its seamless integration of text and images.

在 NCSA 其他人的帮助下，Mosaic 被重写，使其可以在基于 Windows 的机器、Macintosh 以及工作站上运行。作为政府资助实验室的产品，Mosaic 是免费或以象征性的价格提供的。与 UNIX 一样，历史也在重演。但并非完全如此：与 UNIX 的开发者不同，安德森设法迅速将他的发明商业化。1994 年初，硅谷图形公司的创始人吉姆·克拉克 (Jim Clark) 找到他，建议他们将这项发明商业化。安德森同意了，但伊利诺伊大学显然反对这个想法。就像半个世纪前的宾夕法尼亚大学一样，伊利诺伊州看到了校园内所做工作的价值，但却没有看到从事这项工作的人的更大价值。克拉克离开了硅谷图形公司，并于同年春天与安德森一起创立了 Mosaic Communications。伊利诺伊大学声称对 Mosaic 这个名字拥有所有权，因此该公司更名为 Netscape Communications Corporation。克拉克和安德森访问了香槟-厄巴纳，并迅速聘请了许多曾参与该软件开发的程序员。Netscape 于 1994 年 9 月推出了其版本的浏览器。伊利诺伊大学继续提供 Mosaic，并与另一家公司签订了许可协议，但 Netscape 的软件很快取代了 Mosaic，成为该程序最受欢迎的版本^。51

With the help of others at NCSA, Mosaic was rewritten to run on Windows-based machines and Macintoshes as well as workstations. As a product of a government-funded laboratory, Mosaic was made available free or for a nominal charge. As with the UNIX, history was repeating itself. But not entirely: unlike the developers of UNIX, Andreessen managed to commercialize his invention quickly. In early 1994 he was approached by Jim Clark, the founder of Silicon Graphics, who suggested that they commercialize the invention. Andreessen agreed, but apparently the University of Illinois objected to this idea. Like the University of Pennsylvania a half-century before it, Illinois saw the value of the work done on its campus, but it failed to see the much greater value of the people who did that work. Clark left Silicon Graphics, and with Andreessen founded Mosaic Communications that spring. The University of Illinois asserted its claim to the name Mosaic, so the company changed its name to Netscape Communications Corporation. Clark and Andreessen visited Champaign-Urbana and quickly hired many of the programmers who had worked on the software. Netscape introduced its version of the browser in September 1994. The University of Illinois continued to offer Mosaic, in a licensing agreement with another company, but Netscape’s software quickly supplanted Mosaic as the most popular version of the program.⁵¹

1995 年 8 月 8 日，Netscape 公开发行股票。投资者竞相抢购，股价从发行价每股 28 美元飙升至第一天的 58 美元；当晚，网络新闻播报了人们以发行价成功购得股票的故事。公众现在了解到一批新的“亿万富翁”，他们不仅了解了“dot.com”、“HTTP”和“HTML”，还了解到了这些知识。几个月内，Netscape 的股价一度超过每股 150 美元，随后回落。阅读报纸报道和观看电视新闻，人们会觉得 Netscape 上市的那一天标志着计算机历史的真正开始，其他一切都是序幕。对于这个故事来说，这一事件将标志着结束。

On August 8, 1995, Netscape offered shares to the public. Investors bid the stock from its initial offering price of $28 a share to $58 the first day; that evening the network news broadcast stories of people who had managed to get shares at the initial price. The public now learned of a crop of new “instant billionaires,” adding that knowledge to their awareness of “dot.com,” “HTTP,” and “HTML.” Within a few months Netscape shares were trading at over $150 a share, before falling back. Reading the newspaper accounts and watching the television news, one had the feeling that the day Netscape went public marked the real beginning of the history of computing, and that everything else had been a prologue. For this narrative, that event will mark the end.

结论

Conclusion

自 1945 年以来，计算从未停滞不前，20 世纪 90 年代也不例外。互联网的出现是这些年来最大的新闻，尽管这也是办公室台式计算机整合的时期。基于早期发明的 Intel、DOS、Macintosh 和 UNIX 标准，台式计算机达到了稳定状态。大多数办公室使用个人计算机进行文字处理、电子表格和数据库；唯一的新增功能是通过局域网实现的通信。出现了一种新型计算机，称为笔记本电脑（后来，由于重量减轻，称为笔记本），但它们的功能与 PC 相似。事实上，它们被宣传为与桌面上的软件兼容。20 世纪 70 年代后期做出的基本架构决策，包括微处理器的选择和磁盘操作系统的结构，仍然保留了下来（RISC 是一个小而重要的例外）。网络为某些人带来了计算概念上的潜在转变，但截至 1995 年，它还没有取代个人办公桌上的自主通用计算机的概念。随着万维网的成熟，一些人认为消费者所需要的只是一台简单的互联网设备——哑终端的再生——而不是一台通用 PC。但本研究引用的众多例子——IBM 650、1401、PDP-8、Apple II——都支持这一论点，即市场每次都会选择一台物美价廉的通用计算机。

Since 1945 computing has never remained stagnant, and the 1990s were no exception. The emergence of the Internet was the biggest story of these years, although it was also a time of consolidation of the desktop computer in the office. Desktop computing reached a plateau based on the Intel, DOS, Macintosh, and UNIX standards that had been invented earlier. Most offices used personal computers for word processing, spreadsheets, and databases; the only new addition was communications made possible by local-area networking. A new class of computer emerged, called the laptop (later, as it lost more weight, the notebook), but these were functionally similar to PCs. Indeed, they were advertised as being software-compatible with what was on the desk. The basic architectural decisions made in the late 1970s, including the choice of a microprocessor and the structure of a disk operating system, remained (with RISC a small but significant exception). Networking promised for some a potential conceptual shift in computing, but as of 1995 it had not replaced the concept of an autonomous, general-purpose computer on individual desks. As the World Wide Web matures, some argue that all the consumer will need is a simple Internet appliance—a reincarnation of the dumb terminal—not a general-purpose PC. But the numerous examples cited in this study—the IBM 650, the 1401, the PDP-8, the Apple II—all support the argument that the market will choose a good, cheap, general-purpose computer every time.

20 世纪 90 年代最大的新闻是 Altair（一款价值 400 美元的零件套装，曾登上《大众电子》杂志封面）如何成功击败 IBM、王安、UNIVAC、Digital 和 Control Data Corporation 等巨头。IBM 几乎凭借其个人电脑实现了转型，但未能在建立的滩头阵地上继续前进，导致1991 年至 1993 年间损失了数十亿美元^。52个人电脑利润越来越多地流向戴尔、康柏等新公司，尤其是微软。IBM 得以复苏，但这是在其放弃不裁员政策（甚至在整个 20 世纪 30 年代它都坚持这一政策）之后，当它从危机中走出来时，发现微软占据了中心舞台。甚至连计算机协会的伞式贸易组织美国信息处理协会联合会 (AFIPS) 也在 1990 年 12 月 31 日解散^。53

The biggest story of the 1990s was how the Altair, a $400 kit of parts advertised on the cover of Popular Electronics, managed to bring down the mighty houses of IBM, Wang, UNIVAC, Digital, and Control Data Corporation. IBM almost made the transition with its personal computer, but its inability to follow through on the beachhead it established led to multi-billion-dollar losses between 1991 and 1993.⁵² Personal computer profits went increasingly to new companies like Dell, Compaq, and above all Microsoft. IBM recovered, but only after abandoning its no-layoff policy (which it had held to even through the 1930s), and when it emerged from that crisis it found Microsoft occupying center stage. Even the American Federation of Information Processing Societies (AFIPS), the umbrella trade organization of computer societies, perished on December 31, 1990.⁵³

当然，改变计算的不仅仅是 400 美元的 Altair。DEC 和 Data General 也功不可没，但 DEC 和 Data General 都未能在他们打下的基础之上再接再厉。人们可以理解 IBM 的失败，因为它的传统是打孔卡批量处理，并且它不断在法庭上与原告争辩，原告指控它太大。但要理解 Route 128 小型计算机公司为何未能实现转型就不那么容易了。这些公司是处理器和总线架构、紧凑封装、交互式操作和低单位成本方面的先驱。在哈佛商学院的 General Doriot 的领导下，他们也是第一批做后来成为硅谷标志性特征的事情的人：利用风险资本创办一家科技型公司。Netscape 之所以引起公众的极大兴趣，是因为它表明这一传统仍然存在。甚至有可能，这家为了开发以互联网为中心的计算模式而成立的公司，可能能够对微软做出微软刚刚对 DEC、IBM 和其他建立在早期、现已过时的计算模式上的公司所做的事情。

Of course it was not simply the $400 Altair that changed computing. DEC and Data General had a lot to do with that as well, but neither DEC nor Data General were able to build on the foundations they had laid. One could understand IBM’s failings, with its tradition of punched-card batch processing, and its constant courtroom battles against plaintiffs charging that it was too big. It is not as easy to understand how the Route 128 minicomputer companies failed to make the transition. These were the companies that pioneered in processor and bus architectures, compact packaging, interactive operation, and low unit costs. Led by General Doriot of the Harvard Business School, they also were the first to do what later became a defining characteristic of Silicon Valley: to start up a technology-based company with venture capital. Netscape generated so much public interest because it showed that this tradition was still alive. There was even a possibility that this company, founded to exploit a model of computing centered on the Internet, might be able to do to Microsoft what Microsoft had just done to DEC, IBM, and the others who were founded on earlier, now-outdated models of computing.

截至 1995 年，Digital 和 Data General 仍在营业，尽管两家公司都在苦苦挣扎，规模也大幅缩水。Data General 的衰落始于 20 世纪 80 年代初，当时 Tracy Kidder 的《新机器的灵魂》成为第一批登上畅销书榜单的计算机行业书籍之一。该书记录了 Data General 试图追赶 VAX 并重新夺回其在小型计算机领域的领导地位。它抓住了计算机业务的青春、活力和动力，并且仍然是对当今计算机业务的准确描述。然而，由于缺乏我们现在都拥有的 20-20 后见之明，Kidder 并没有提到 Data General 的 Nova（“清洁机器”）如何启发了个人电脑设计师，包括 Ed Roberts 和 Steve Wozniak。Data General 的某个人可能建议了另一种做法：忽略 VAX，而是专注于它帮助实现的小型系统。如果是这样，Kidder 的书并没有记录它。

As of 1995 Digital and Data General were still in business, although both were struggling and much-reduced in size. Data General’s decline began in the early 1980s, just when Tracy Kidder’s The Soul of a New Machine became one of the first books about the computer industry to get on the best-seller list. That book chronicled Data General’s attempt to chase after the VAX and regain its leadership in minicomputers. It captured the youth, energy, and drive of the computer business, and it remains an accurate description of the computer business today. Lacking the 20-20 hindsight that we now all have, Kidder did not, however, mention how Data General’s Nova, the “clean machine,” had inspired the designers of personal computers, including Ed Roberts and Steve Wozniak. Someone at Data General may have recommended an alternate course: that it ignore the VAX and concentrate instead on the small systems it had helped bring into being. If so, Kidder’s book does not record it.

1992 年，肯·奥尔森辞去了数字设备公司负责人一职，因为他创立的公司正走向破产。新闻报道中经常将奥尔森和疲惫不堪的 DEC 与年轻的比尔·盖茨和他生机勃勃的微软进行对比。但很少有人意识到这种对比的讽刺意味。盖茨在 PDP-10 上学会了编程，而 DEC 对微软软件的影响也十分显著。不仅如此，数字设备公司还推动了微软等公司的发展。有人说，如果没有奥尔森，我们现在还在用打孔卡编程。这听起来像是出于同情而做出的慷慨夸大之词；事实上，我们可以将奥尔森的功绩归功于奥尔森，甚至更多。现代计算是 JCR Licklider 所说的“人机共生”愿景如何实现的故事。这是利克莱德本人以及道格·恩格尔巴特、泰德·霍夫、埃德·罗伯茨、史蒂夫·乔布斯、史蒂夫·沃兹尼亚克、比尔·盖茨、加里·基尔达尔、蒂姆·伯纳斯-李等人的努力的成果。在这份名单中，也许应该加上肯·奥尔森这个名字。对资本主义制度的“创造性破坏”创造了奇迹，但这一过程既不合理也不公平。

In 1992, Ken Olsen resigned as head of Digital, as the company he founded was heading toward bankruptcy. A typical news story contrasted Olsen and a tired DEC with the young Bill Gates and his vibrant Microsoft. Few saw the irony of that comparison. Gates learned how to program on a PDP-10, and we have seen DEC’s influence on Microsoft’s software. More than that: Digital Equipment Corporation set in motion the forces that made companies like Microsoft possible. One person was quoted stating that were it not for Olsen we would still be programming with punched cards. That sounded like a generous overstatement made out of sympathy; in fact, one could credit him with doing that and much more. Modern computing is a story of how a vision of “man-machine symbiosis,” in J. C. R. Licklider’s term, came to fruition. That happened through the efforts of people like Licklider himself, as well as Doug Engelbart, Ted Hoff, Ed Roberts, Steve Jobs, Steve Wozniak, Bill Gates, Gary Kildall, Tim Berners-Lee, and many others. To that list, perhaps near the top, should be added the name Ken Olsen. The “creative destruction” of the capitalist system had worked wonders, but the process was neither rational nor fair.

“互联网时代”，1995-2001年

“Internet Time,” 1995–2001

第一版的叙述结束于 1995 年 8 月 8 日，即 Netscape 在股票市场上市的那一天。互联网的商业化以及 Netscape 在其中扮演的角色开创了计算的新时代。现在撰写这个时代的历史还为时过早。目前还没有明确的理论框架供历史学家构建叙述。不过，过去几年发生了这么多事情，人们不能推迟尝试撰写有关“.com”现象的历史叙述。本章将用一节来做这件事，但这篇关于互联网泡沫膨胀和破灭的编年史仍然在进行中。

The narrative in the first edition ended on August 8, 1995, the day that Netscape offered shares on the stock market. The commercialization of the Internet, and the role that Netscape played in it, ushered in a new era in computing. It is too early to write a history of this era. There is no clear theoretical framework on which the historian can build a narrative. Still, so much has happened in the past few years that one cannot put off an attempt to write some kind of historical narrative about the “dot.com” phenomenon. A section of this chapter will do that, but this chronicle of the inflation and bursting of the dot.com bubble is very much a work in progress.

本章还讨论了过去几年的另外两个发展。就像互联网现象一样，如果你读报纸头条，就会发现这些发展的方向似乎每天都在变化。幸运的是，这些发展与计算机“古代历史”（即 1995 年之前）的事件有着很好的联系。因此，它们让历史学家获得了一丝洞察力。首先讨论的针对微软的反垄断审判是针对该公司的一系列法律行动的高潮，它反映了微软早在 1975 年成立时就存在的问题。不仅如此，微软的审判还呼应了 1970 年代 IBM 与美国司法部发生法律纠纷时提出的许多针对 IBM 的论点。

This chapter also addresses two other developments of the past few years. Like the dot.com phenomenon, these are ongoing developments whose direction seems to change daily if one reads the newspaper headlines. Fortunately, these developments have nice connections to events of computing’s “ancient history” (i.e., before 1995). Thus they allow the historian to gain a glimmer of perspective. The antitrust trial against Microsoft, discussed first, is the culmination of a sequence of legal actions taken against the company, and it reflects issues that were present at Microsoft as early as 1975, when the company was founded. Not only that, the Microsoft trial echoes many of the arguments made against IBM during its legal troubles with the U.S. Justice Department in the 1970s.

最后讨论的 GNU/Linux 操作系统和“开源”软件运动同样有着深厚的根源。第 3 章讨论了 SHARE 的创立，以及关于谁有权使用和修改 TRAC 编程语言的争议。GNU/Linux 是 UNIX 的一个变体，UNIX 是 20 世纪 60 年代末开发的系统，本书前面的几章对其进行了详细讨论。UNIX 几乎从一开始就是一个开放系统，尽管与现在的“开放”定义并不完全一样。与针对微软的反垄断审判一样，开源软件运动与个人电脑发明的初期也有着密切的联系。微软及其创始人的早期行动在这里也发挥了重要作用。我们从反垄断审判开始。

The discussion of the GNU/Linux operating system and the “open source” software movement, discussed last, likewise has deep roots. chapter 3 discussed the founding of SHARE, as well as the controversy over who was allowed to use and modify the TRAC programming language. GNU/Linux is a variant of UNIX, a system developed in the late 1960s and discussed at length in several earlier chapters of this book. UNIX was an open system almost from the start, although not quite in the ways that “open” is defined now. As with the antitrust trial against Microsoft, the open source software movement has a strong tie to the beginnings of the personal computer’s invention. Early actions by Microsoft and its founders played an important role here as well. We begin with the antitrust trial.

微软

Microsoft

在比赛第三节播出的广告是 1984 年 1 月超级碗赛事转播中最令人难忘的部分（见第 8 章）。苹果向我们保证，Macintosh 将开启个人计算的新纪元，因此 1984 年将不会像乔治奥威尔的小说《1984》所预言的那样，是沉闷的墨守成规和压迫的一年。个人计算革命确实正在酝酿之中，而 Macintosh 正引领潮流。但帮助将这场革命推向大众市场的是微软，而不是苹果。那不是 1984 年（Mac 问世的那一年），而是 1992 年，当时微软开始发布其 Windows 程序 3.1 版。1984 年，苹果希望 Mac 能将施乐帕洛阿尔托研究中心的创新理念（已经存在于一些个人计算机系统中）带给消费者。十几年后，微软（而不是苹果）将主宰个人计算机软件。¹而这种主导地位又会导致其陷入激烈的反垄断审判。

A commercial, aired during the third quarter of the game, was the most memorable part of the broadcast of the January 1984 Super Bowl (see chapter 8). The Macintosh, Apple assured us, would usher in a new era of personal computing, and therefore the year 1984 would not be one of dreary conformity and oppression as prophesied by George Orwell’s novel 1984. A revolution in personal computing was indeed in the works, and the Macintosh was leading the way. But Microsoft, not Apple, helped bring the revolution to a mass market. That happened not in 1984, the year the Mac appeared, but in 1992, when Microsoft began shipping version 3.1 of its Windows program. In 1984, Apple hoped that the Mac would bring the innovative ideas from the Xerox Palo Alto Research Center, ideas already present in a few personal computer systems, to the consumer. A dozen years later, Microsoft, not Apple, would dominate personal computer software.¹ And that domination, in turn, would lead to its entanglement in a bitter antitrust trial.

正如 IBM 在 20 世纪 70 年代花费了相当一部分资源来应对美国司法部的挑战一样，微软在 1997 年也面临同样的境地，当时它被起诉。2001 年 11 月，联邦政府宣布和解，但几个州和欧盟拒绝接受。他们的论点也被 2002 年 11 月 1 日的裁决驳回。几乎每天都有商业媒体报道法官或律师发表声明。在案件解决之前，人们只能对其重要性做出初步评论。然而，IBM 审判的教训适用于本案对微软的诉讼：即司法部并不是一个认识到技术进步将如何使大部分诉讼变得无关紧要的地方。在 IBM 审判期间，当诉讼双方争夺大型机主导地位时，个人电脑的微软对应物是什么？现在下结论还为时过早，尽管我将在本章后面讨论一些候选者。可以肯定的是，计算技术的进步已经威胁到、并将继续威胁微软基于其 Windows 和 Office 软件主导个人计算的能力。

Just as IBM spent a significant fraction of its resources during the 1970s facing a challenge by the U.S. Justice Department, so too is Microsoft in the same situation, following a similar filing against it in 1997. In November 2001 the federal government announced a settlement, but several states, and the European Union, refused to go along. Their arguments were also rejected by a ruling on November 1, 2002. Almost daily, the business press reports whenever a judge or lawyer makes a statement. Until the case is settled, one can only make provisional comments about its significance. The lesson of the IBM trial, however, applies to the present case against Microsoft: namely that the Justice Department is not a place that recognizes how advancing technology will render much of the lawsuit irrelevant. What is the Microsoft-equivalent of the personal computer, whose appearance in the midst of the IBM trial was ignored as the litigants fought over mainframe dominance? It is too early to tell, although I will discuss some candidates later in this chapter. What is certain is that advances in computing already threaten, and will continue to threaten, Microsoft’s ability to dominate personal computing, based on its Windows and Office software.

微软和英特尔的授权政策催生了戴尔、康柏和 Gateway 等克隆制造商，它们为苹果客户提供了无法获得的选择。（在苹果的大部分历史中，它一直拒绝向第三方电脑制造商授权其 Macintosh 软件。）这项政策带来了更多种类的产品，最重要的是，降低了基于英特尔微处理器并运行微软 DOS 和 Windows 的电脑的价格。Windows 3.1 版、英特尔推出的奔腾处理器以及微软将应用软件整合成名为 Microsoft Office 的套件，这些因素结合起来，让消费者能够以更低的价格获得 Macintosh 提供的 80% 的功能。令苹果惊讶的是（也让 Mac 粉丝感到懊恼），这个百分比足以让苹果永远失去优势。到 1995 年，苹果更优雅的设计优势不再重要，因为微软/英特尔的组合成为了标准，就像 20 世纪 60 年代的 COBOL 一样。与 COBOL 一样，重要的是标准的存在，而不是软件的内在价值或缺乏价值。

The licensing policies of Microsoft and Intel gave rise to clone manufacturers, like Dell, Compaq, and Gateway, who provided choices unavailable to Apple customers. (Apple, for most of its history, has refused to license its Macintosh software to third-party computer makers.) That policy yielded a greater variety of products and, above all, lower prices for computers based on Intel microprocessors and running Microsoft’s DOS and then Windows. Windows version 3.1, Intel’s introduction of the Pentium processor, and Microsoft’s combining applications software into a suite called Microsoft Office, combined to give consumers, let’s say, 80 percent of what the Macintosh was offering, at a lower price for the total package. To Apple’s surprise (and to the chagrin of Mac fans), that percentage was good enough to tip the balance, perhaps forever, away from Apple. By 1995 the advantage of Apple’s more elegant design no longer mattered, as the Microsoft/Intel combination became a standard, like COBOL in the 1960s. As with COBOL, what mattered was the very existence of a standard, not the intrinsic value or lack thereof of the software.

Macintosh 连接

The Macintosh Connection

微软的成功和失败可以从很多地方开始，但 Mac 的推出让我们很容易地发现了几个关键因素。首先，当 Mac 于 1984 年问世时，它拥有出色的用户界面，但几乎没有应用软件——人们购买个人电脑时使用的程序。它最有趣的应用程序是 MacPaint，这是一款绘图程序，源自施乐的开创性工作，是 IBM 兼容机上任何软件都无法比拟的。但对于文字处理，这是任何严肃的新电脑都必须具备的应用程序，苹果只提供了 MacWrite，它利用了它的图形界面，但其他方面功能极其有限。2 ^MacPaint和 MacWrite 都是内部开发的。

One could begin this story of Microsoft’s triumph and troubles at any number of places, but the introduction of the Mac conveniently allows us to identify several critical factors. The first was that when the Mac appeared in 1984, it had a magnificent user interface but almost no applications software—the programs that people actually bought personal computers for. The most interesting application that it did have was MacPaint, a drawing program descended from the pioneering work at Xerox, and something that no software for IBM compatibles could approach. But for word processing, an application that any serious new computer had to have, Apple offered only MacWrite, which took advantage of its graphical interface, but which otherwise was extremely limited in capability.² Both MacPaint and MacWrite were developed in-house.

除了这些程序之外，早期的 Mac 用户还可以获得一个电子表格：Multiplan，它由微软为其他计算机开发，但移植到了 Mac 上。尽管一些流行的报道喜欢将比尔盖茨和史蒂夫乔布斯描绘成死对头，但在这一时期的大部分时间里，两人保持着友好和互惠互利的业务关系。在 Mac 开发之初，1981 年 6 月，乔布斯和杰夫拉斯金（Macintosh 的最初构想者）与盖茨会面，次年 1 月，微软同意为新机器开发软件^。3

Besides those programs, early Mac customers could also get a spreadsheet: Multiplan, developed by Microsoft for other computers but ported to the Mac. Although some popular accounts enjoy setting up Bill Gates and Steve Jobs as mortal enemies, for much of this period the two men had a cordial and mutually beneficial business relationship. At the onset of the Mac’s development, in June 1981, Jobs and Jef Raskin (who had the initial idea for the Macintosh) met with Gates, and in January of the following year Microsoft agreed to develop software for the new machine.³

^{盖茨几乎不需要说服就能知道个人电脑的发展方向。1980 年，当微软正在谈判为 IBM PC 提供 DOS 时，盖茨聘请了一位程序员，带领公司走向了相反的方向。这个人就是查尔斯·西蒙尼 (Charles Simonyi)，匈牙利人，他先是在苏联制造的真空管计算机 Ural-II 上学会了编程，然后在安装了高级 ALGOL 编译器的丹麦晶体管计算机上学会了编程。20 世纪 70 年代，西蒙尼在施乐帕罗奥多研究中心工作，在那里他为 Alto 工作站开发了一款名为“Bravo”的文字处理器。Bravo}通常被认为是第一个真正的所见即所得 (WYSIWYG) 显示屏，施乐的其他员工将这一概念带到了苹果。4

Gates needed little convincing of where personal computing was going. Even as Microsoft was negotiating to supply DOS for the IBM PC in 1980, Gates hired a programmer who would take the company in the opposite direction. That was Charles Simonyi, a native of Hungary who learned how to program first on a Soviet-built vacuum-tube computer called the Ural-II, then on a Danish transistorized computer that had an advanced ALGOL compiler installed on it. In the 1970s Simonyi worked at Xerox-PARC, where he developed a word processor called “Bravo” for the Alto workstation. Bravo is often credited with having the first true WYSIWYG (“What-You-See-Is-What-You-Get”) display, a concept that other Xerox employees brought with them to Apple.⁴

1985 年，微软为 Macintosh 制作了另一款电子表格 Excel，它充分利用了 Macintosh 界面提供的所有功能。Excel 大获成功，帮助苹果度过了 Mac 销售面临完全枯竭的困难时期。Mac 用户终于拥有了一款可与 IBM PC 上的 Lotus 1-2-3 相媲美的电子表格。微软的努力也获得了一些收获：除了赢得商业成功外，微软程序员还学会了如何为基于 Windows 的界面开发软件——这是 Lotus 和 Word Perfect 很难学会的。

In 1985 Microsoft produced another spreadsheet, Excel, for the Macintosh, which took advantage of all that the Macintosh interface had to offer. Excel was a success and helped Apple get through a difficult period when Mac sales were in danger of completely drying up. Mac users finally had a spreadsheet that was comparable to Lotus 1-2-3 on the IBM PCs. For its efforts, Microsoft gained something too: besides winning a commercial success, Microsoft programmers learned how to develop software for a Windows-based interface—something that Lotus and Word Perfect would have a hard time learning.

聘用西蒙尼以及微软和苹果之间的这些互动最终导致比尔·盖茨决定在英特尔 80686 平台上重现 Macintosh 体验。考虑一下这个决定的背景。在 20 世纪 80 年代中期，“Windows”只是为 IBM 兼容机提出的众多图形系统（例如 VisiOn、GEM 等）之一。微软的应用程序，如 Multiplan，并不像 Lotus 1-2-3 或 Word Perfect 等程序那样受到行业评论家的好评。Windows 模式也受到一个主要来自 Lotus 的竞争理念的挑战：一个在 DOS 下运行的单一程序，结合了电子表格、数据库和文字处理。Lotus 为 IBM PC 提供了一个名为 Symphony 的程序，并正在为 Mac 开发一个名为 Jazz 的程序。在领先的 PC 数据库软件供应商 Ashton-Tate，一位名叫罗伯特·卡尔的施乐-帕洛阿尔托研究中心校友正在开发一个名为 Framework 的类似程序^。5

The ultimate impact of hiring Simonyi, and of these interactions between Microsoft and Apple, was that Bill Gates decided to recreate the Macintosh experience on the Intel 80686 platform. Consider the context of that decision. In the mid-1980s, “Windows” was but one of many graphical systems (e.g., VisiOn, GEM, et al.) proposed for IBM compatibles. And Microsoft’s applications programs, like Multiplan, were not as well regarded by industry critics as programs like Lotus 1-2-3 or Word Perfect. The Windows model was also being challenged by a competing idea, mainly from Lotus: that of a single program, running under DOS, that combined spreadsheets, databases, and word processing. Lotus offered such a program called Symphony for the IBM PC and was working on one for the Mac called Jazz. At Ashton-Tate, the leading supplier of database software for the PC, a Xerox-PARC alumnus named Robert Carr was developing a similar program called Framework.⁵

事实证明，将应用程序分开，同时要求每个应用程序都遵循通用图形界面的做法将占上风。6^这正是乔布斯对所有 Macintosh 开发人员的要求，而盖茨则将其作为微软的重点（略微不那么尖锐）。西蒙尼开发了一种编程系统，使微软能够随着公司的发展管理越来越大、越来越复杂的编程工作。这种风格涉及一种命名变量的方式，被称为“匈牙利语”，这是一个内部笑话，指的是任何不熟悉微软编程的人都无法理解它，就像西蒙尼的母语匈牙利语对于其他欧洲语言的使用者来说也是如此^。7

It turned out that the practice of keeping the applications separate, while requiring that each adhere to a common graphical interface, would prevail.⁶ That was what Jobs insisted on for all Macintosh developers, and Gates made it the focus (slightly less sharp) at Microsoft. Simonyi developed a system of programming that allowed Microsoft to manage increasingly larger and more complex programming jobs as the company grew. The style involved a way of naming variables, and was called “Hungarian,” an inside joke referring to its incomprehensibility to anyone not familiar with Microsoft’s programming, like Simonyi’s native Hungarian language supposedly is to speakers of other European languages.⁷

“匈牙利式”也许不是关键因素，但不知何故，微软的管理人员学会了如何管理由越来越大的程序员团队编写的复杂软件的开发和引入。还有一种技术尤其具有创新性。尽管这种技术是在其他地方开发的，但微软采用了这种技术，并将其大规模应用，这是其他地方所未见的，它彻底打破了大型机软件公司管理大型项目的方式。在微软，每天结束时，从事新产品某一部分的程序员必须将他们的工作提交到一个中央文件，在那里，它会在晚上与其他所有人的工作一起被编译成每日的“构建”。8^如果你的贡献导致中央文件崩溃，你就有责任修复它。该构建将成为第二天工作的基础。9^更重要的是，一旦构建开始发挥作用，编程团队成员就必须使用它，无论这样做效率有多低。这一要求让生活变得困难，尤其是在软件处于早期阶段且大部分功能都不太好的时候，但它让程序员专注于交付高质量的成品。这个过程也有一个令人回味的名字：“吃自己的狗粮”。10 从那时起，^公众开始意识到微软程序员积累的巨额财富，他们在微软工作的时间足够长，以至于他们的股票期权都归属微软。鲜为人知的是，这些程序员的生活是“狗粮”：不睡觉、吃自动售货机里的食物、无休止地盯着电脑屏幕、没有社交或家庭生活，以及其他磨难，这些磨难让程序员陷入“死亡行军”中，在按时完成程序的同时修复“损坏”的构建^。11

“Hungarian” may not have been the crucial factor, but somehow Microsoft’s managers learned to manage the development and introduction of complex software written by ever-larger teams of programmers. One other technique was especially innovative. Although it had been developed elsewhere, Microsoft embraced this technique and applied it on a large scale not seen elsewhere, and broke radically from the way large projects were managed at mainframe software houses. At Microsoft, programmers working on a section of a new product were required to submit their work to a central file at the end of each day, where overnight it would be compiled, along with everyone else’s, into a daily “build.”⁸ If your contribution caused the central file to crash, you were responsible for fixing it. That build then became the basis for the next day’s work.⁹ What was more, as soon as the build became marginally functional, members of the programming team were required to use it, regardless of how inefficient that might be. This requirement made life difficult, especially when the software was in an early stage and little of it worked well, but it kept the programmers focused on shipping a finished product of high quality. This process, too, had an evocative name: “eating your own dog food.”¹⁰ The public has since become aware of the large fortunes amassed by Microsoft programmers who worked there long enough to have their stock options vest. Less well known is the dog’s life of no sleep, eating out of vending machines, endless hours spent staring into a computer screen, no social or family life, and other tribulations for a programmer caught in the “death march” of fixing a “broken” build while getting a program finished on time.¹¹

这些努力的累积效应是 Windows 版本不断改进，应用程序套件也不断扩展。Word 和 Excel 是应用软件的两大支柱，很快数据库程序 Access、演示程序 PowerPoint、项目管理程序 Project 和许多其他程序也加入其中（表 10.1）。微软从较小的独立公司购买了许多这些程序，然后对其进行重新开发以符合 Windows 界面。12^主要应用程序被合并到名为 Microsoft Office 的应用程序套件中。13^累积效应改变了微软的收入来源。在早期，微软主要销售 BASIC 和其他语言编译器；1981 年以后，它的收入主要来自个人电脑的 DOS 操作系统。到 1991 年，微软 50% 以上的收入来自应用程序，尤其是 Office。Windows 和 Office 的强大影响力压倒了操作系统供应商中的 IBM 和 Digital Research，以及应用程序提供商中的 Lotus、Ashton-Tate 和 Word Perfect。到了 20 世纪 90 年代中期，许多独立软件公司为 Windows 平台提供应用程序，但只有少数公司规模较大，而提供文字处理、数据库或电子表格应用程序的公司则更少。

The cumulative effect of these efforts was a steady stream of ever-improved versions of Windows and an ever-expanding suite of applications. Word and Excel were the two pillars of applications software, soon joined by the database program Access, the presentation program PowerPoint, the project management program Project, and a host of others (table 10.1). Microsoft purchased many of these programs from smaller, independent companies and then reworked them to conform to the Windows interface.¹² Major applications were combined into an applications suite called Microsoft Office.¹³ The cumulative effect was to change the revenue stream for Microsoft. In its earliest days Microsoft mainly sold BASIC and other language compilers; after 1981 it derived its revenues primarily from the DOS operating system for the PC. By 1991 over 50 percent of Microsoft’s revenues came from applications, especially Office. The resulting juggernaut of Windows and Office rolled over IBM and Digital Research among the operating system suppliers, and Lotus, Ashton-Tate, and Word Perfect among the applications providers. By the mid-1990s, many independent software companies supplied applications for the Windows platform, but only a few were of significant size, and fewer still offered word processing, database, or spreadsheet applications.

表 10.1

1983 年至 2001 年微软软件年表

Table 10.1

Selected chronology of Microsoft Software, 1983–2001

年	Windows 版本	应用	英特尔处理器
1983	“界面管理器”已发布，但尚未发布	个人电脑版 Word（DOS）
1984		项目；图表；Macintosh 版 Word	80286
1985	1.0	适用于 PC (DOS) 的 Word 2.0；适用于 Macintosh 的 Excel
1986		作品	80386
1987	2.0	Forefront（后来的 PowerPoint）；PC 版 Excel（Windows）
1988
1989			80486
1990	3.0	适用于 Windows 1.0 的 Word；Office
1991
1992	3.1	使用权
1993	3.11；新约	办公室 4.0	奔腾
1994	新台币 3.5
1995	95（《芝加哥》）	Office 95；网络 (MSN)；Internet Explorer 1.0	奔腾专业版
1996		Internet Explorer 3.0；Exchange
1997		Office 97；Internet Explorer 4.0	MMX；奔腾 2
1998	98	Hotmail	赛扬
1999			奔腾3
2000	2000；我	办公室 2000
2001	经验值		奔腾4

（来源：数据来自多个来源，包括 Michael A. Cusumano 和 Richard W. Selby 所著的《微软的秘密》（纽约：自由出版社，1995 年）；Martin Campbell-Kelly 所著的《不仅仅是微软：个人电脑软件产业的成熟，1982-1995 年》，《商业史评论》（2001 年春季）：103-145 页；Stephen Manes 和 Paul Andrews所著的《盖茨：微软大亨如何重塑一个产业并成为美国首富》（纽约：Doubleday，1993 年）。资料来源还包括 1990 年至 1994 年《Infoworld》和《PC Week》精选文章作者所拥有的剪报文件。在某些情况下，日期是近似值，反映了产品发布和实际提供给消费者之间的差异。）

(Source : Data taken from a number of sources, including Michael A. Cusumano and Richard W. Selby, Microsoft Secrets (New York: Free Press, 1995); Martin Campbell-Kelly, “Not Only Microsoft: The Maturing of the Personal Computer Software Industry, 1982–1995,” Business History Review (Spring 2001): 103–145; Stephen Manes and Paul Andrews, Gates: How Microsoft’s Mogul Reinvented an Industry, and Made Himself the Richest Man in America (New York: Doubleday, 1993). Sources also include a clipping file in the possession of the author of selected articles from Infoworld and PC Week, 1990–1994. In some cases the dates are approximate, reflecting the difference between the announcement of a product and its actual availability to consumers.)

IE浏览器

Internet Explorer

当这个庞然大物最终引起美国司法部反垄断律师的注意时，微软以外的个人电脑行业观察家们几乎没有人感到惊讶，因为微软的竞争对手们的压力一直在增加。引发诉讼的具体行动是微软将 Web 浏览器捆绑到 Windows 中。1994 年 12 月，微软向 Spyglass 支付了许可费，以将其工作用作 Web 浏览器的基础，微软将其更名为 Internet Explorer。（Spyglass 和 Netscape 一样，都是伊利诺伊大学 Mosaic 的后代。）1995 年夏天，就在 Netscape 公开募股后不久，微软将 Spyglass 浏览器的一个版本作为 Windows 的一部分提供。从这一点开始，微软走上了一条熟悉的道路：它发布了浏览器的连续版本，每个版本都具有更多功能并与基本操作系统的功能进行了更多集成。请注意，通过将 Internet Explorer 捆绑到 Windows 中并以单一价格出售，微软有效地阻止了 Spyglass 对其浏览器的 Windows 版本收取费用，至少在零售方面是这样。

When that juggernaut finally caught the attention of antitrust lawyers at the U.S. Justice Department, few observers of the personal computer industry outside of Microsoft were surprised, because pressure had been building up among Microsoft’s competitors. The specific action that triggered the lawsuit was Microsoft’s bundling a Web browser into Windows. In December 1994, Microsoft paid Spyglass for a license to use its work as the basis for a Web browser, which Microsoft renamed Internet Explorer. (Spyglass, like Netscape, descended from Mosaic at the University of Illinois.) In the summer of 1995, just after Netscape’s public offering, Microsoft offered a version of Spyglass’s browser as part of Windows. From this point Microsoft followed a familiar road: it issued successive versions of the browser, each one with more features and more integration with the base operating system’s functions. Note that by bundling Internet Explorer into Windows and selling it at a single price, Microsoft effectively prevented Spyglass from charging money, at least for retail sales, for a Windows version of its browser.

1997 年秋季推出的 Internet Explorer 4.0 对微软来说只不过是另一个新版本。对于 Netscape 和司法部来说，这完全是另一回事。在他们看来，微软与 IE 4.0 的紧密集成是违反反垄断法的行为。之前曾有人警告过麻烦。由于事态发展迅速，到 1997 年，很少有人记得这些警告。下面简要概述了最明显的警告。还介绍了与这些行动相关的一些有趣短语。

Internet Explorer 4.0, introduced in the fall of 1997, was just another new version to Microsoft. It was something else entirely to Netscape and to the Justice Department. In their view, Microsoft’s tight integration of IE 4.0 was an action contrary to antitrust laws. There had been earlier warnings of trouble. With the fast pace of events, few remembered these warnings by 1997. What follows is a brief overview of only the most visible warnings. Some of the colorful phrases that arose in connection with those actions are also introduced.

第一个法律问题的迹象并不涉及微软，但却引入了一个在后来的审判中出现的短语。这是 1987 年 Lotus 起诉 Paperback Software 公司的一桩诉讼，Paperback Software 是由便携式电脑名人 Adam Osborne 创立的。Paperback 销售的电子表格功能与 1-2-3 完全相同，但价格却只有后者的一小部分。14 ^Lotus指控 Paperback 即使没有抄袭或窃取 Lotus 的代码，也抄袭了 1-2-3 的“外观和感觉”，这是违法的。在诉讼进行期间，1988 年苹果起诉微软（和惠普）抄袭了 Windows 2.0 版中 Macintosh 的“外观和感觉”。苹果和微软签署了许可协议，但苹果指控它只授权了 Windows 1.0 的 Macintosh 界面。值得注意的是，当时苹果的负责人是约翰·斯卡利，而不是史蒂夫·乔布斯。乔布斯不仅承认，甚至还吹嘘自己从施乐-帕洛阿尔托研究中心窃取了图形界面。15^整个1989 年，这起案件一直拖延，最终被各种事件所取代。双方也意识到，他们之间有做生意的根本需要，这种需要可以追溯到 20 世纪 70 年代两家公司成立之初^。16

The first indication of a legal issue did not involve Microsoft but did introduce a phrase that would figure in later trials. This was a suit filed in 1987 by Lotus against a company called Paperback Software, established by Adam Osborne of portable computer fame. Paperback was selling a spreadsheet that functioned identically to 1-2-3, but at a fraction of the price.¹⁴ Lotus charged that Paperback, even if it did not copy or steal Lotus’s code, nonetheless copied the “look and feel” of 1-2-3, and that was illegal. While that lawsuit was in progress, in 1988 Apple sued Microsoft (and Hewlett-Packard) for copying the “look and feel” of the Macintosh in version 2.0 of Windows. Apple and Microsoft had signed a licensing agreement, but Apple charged that it had only licensed the Macintosh interface for Windows 1.0. It is worth noting that by this time the head of Apple was John Sculley, not Steve Jobs. Jobs not only admitted but even boasted of having stolen the graphical interface from Xerox-PARC.¹⁵ Throughout 1989 the case dragged on, eventually to be overtaken by events. Both parties also realized that they had a fundamental need to do business with one another, a need that went back to the founding of both companies in the 1970s.¹⁶

1990 年，联邦贸易委员会调查了微软与 IBM 就联合开发 IBM/Microsoft 操作系统达成的协议，IBM 将该操作系统作为 OS/2 进行营销。联邦贸易委员会还调查了一项指控，即 1990 年，微软获得了一家名为 GO 的初创公司开发的笔式计算机原型的详细信息，然后在一次贸易展上宣布它将很快将笔式输入集成到 Windows 中（微软从未真正做过这件事）。这导致 GO 的所有财务支持立即枯竭，最终倒闭。17^这种手法被称为“恐惧、不确定和怀疑”，简称“FUD”。20 世纪 60 年代，当 IBM 宣布推出其 System/360（型号 91）与 Control Data 的超级计算机竞争时，Control Data 出于同样的原因对 IBM 提出了这一指控（第 5 章）。当占主导地位的供应商承诺同样的技术将很快成为其主流产品线的一部分时，谁会购买一家小公司的产品呢？¹⁸当时还出现了另一个说法，即微软对 GO 采取的行动相当于“切断其空气供应”：使 GO 无法销售其产品或筹集资金。一些人指责微软在捆绑 Internet Explorer 时也对 Spyglass 采取了同样的措施。从 1999 年到 2001 年，诉讼当事人、记者和法官都会对这句话进行长篇大论的分析，看它是否适用于 Netscape。

In 1990 the Federal Trade Commission investigated Microsoft in connection with its agreements with IBM over the development of a joint IBM/Microsoft operating system, which IBM marketed as OS/2. The FTC had also investigated a charge that in 1990, Microsoft gained access to details of a prototype pen-based computer developed by a startup called GO, and then announced at a trade show that it would soon integrate pen-based input into Windows (something Microsoft never really did). The effect was to immediately dry up all financial support for GO, which eventually folded.¹⁷ This technique was known by the phrase “Fear, Uncertainty, and Doubt,” or “FUD.” It was a charge that Control Data leveled against IBM in the 1960s for the same reason, when IBM announced its System/360, Model 91, to compete with Control Data’s supercomputer (chapter 5). Who would buy a small company’s product when the dominant vendor promised that the same technology would soon be part of its mainstream line?¹⁸ Another phrase, which emerged at this time, was that Microsoft’s actions against GO amounted to “cutting off its air supply”: making it impossible for GO to sell its product or to raise money. Some accused Microsoft of doing that to Spyglass as well, when it bundled Internet Explorer. Through 1999 and into 2001, litigants, journalists, and judges alike would parse this phrase at great length, as it applied—or not—to Netscape.

1994 年，美国司法部曾威胁要起诉微软，指控其将产品捆绑到 Windows 中，但在微软签署了一项同意令后，司法部撤诉。微软承诺不会将一种产品与另一种产品“捆绑销售”，即坚持要求购买 Windows 的客户也购买另一种微软产品。这种捆绑销售的概念众所周知，其历史根源在于 20 世纪早期的反垄断立法。尽管微软小心翼翼地保护 Windows 的源代码，但它同意开放与应用程序交互的 Windows 代码部分：即所谓的应用程序接口 (API)。因此，例如，数据库程序的开发人员可以放心，他们的产品在理论上可以像微软开发的任何数据库一样与 Windows 顺利运行。

The Justice Department had threatened to sue Microsoft in 1994 over bundling of products into Windows, but it dropped the suit after Microsoft entered into a Consent Decree. Microsoft promised not to engage in the practice of “tying” sales of one product to another—that is, insisting that customers who bought Windows also buy another Microsoft product. This concept of the tie-in was well understood and had historical roots in early-twentieth-century antitrust legislation. Although Microsoft carefully guarded the source code for Windows, it agreed to make available the parts of Windows code that interacted with applications programs: the so-called Applications Program Interface, or API. Thus, for example, developers of a database program were assured that their product would, in theory, work as smoothly with Windows as any database developed by Microsoft would.

然而，该公司制定的 Windows 收费政策却不那么慷慨。到 1995 年，消费者很少购买“收缩包装”的 Windows 并自行安装；相反，他们购买一台出厂时已安装 Windows 的计算机。这使得本来就很低的分销成本变得更低；计算机公司可以协商降低 Windows 成本，并将节省的成本转嫁给消费者；消费者无需为繁琐的安装过程而烦恼。根据《同意令》，微软不能坚持要求任何从 Compaq 购买计算机的人也必须购买 Windows。然而，微软是按照“每个处理器”而不是实际安装的 Windows 程序数量向 Compaq 收费的。因此，即使 Compaq 出售的计算机安装了另一个操作系统（甚至没有操作系统），也必须为 Windows 付费。这是合法的，但这项政策并不能平息越来越多的微软批评者^。19

However, the company developed a policy in the way it charged for Windows that was less magnanimous. By 1995 consumers rarely bought Windows in a “shrink-wrapped” package and installed it themselves; instead they bought a computer on which Windows was already installed at the factory. That brought distribution costs, already low, even lower; the computer companies could negotiate for a low cost for Windows and could pass on the savings; and the consumer did not need to bother with a cumbersome installation process. By the Consent Decree, Microsoft could not insist that anyone who bought a computer from, say, Compaq had to buy Windows, too. However, Microsoft billed Compaq on a “per processor” basis, not on the actual numbers of Windows programs installed. Therefore Compaq had to pay for a copy of Windows even if it sold a computer that had another operating system—even no operating system—installed on it. Legal, but not a policy to calm the growing army of Microsoft critics.¹⁹

1995 年还发生了一起事件，在行业媒体上引起了不小的波澜，但事后看来，这似乎进一步激怒了微软的竞争对手和司法部的人。那一年，微软宣布将收购金融软件 Quicken 的制造商 Intuit，该公司是少数几个拥有主导市场份额的应用程序独立供应商之一。在微软发起收购后（批评者指责，微软是在学习了 Intuit 最优秀程序员的技术后），由于司法部的反对，收购被取消。显然，反对微软的情绪正在高涨。

One more event occurred in 1995 that caused a minor ripple in the trade press, but in hindsight it appears to have further enraged Microsoft’s competitors and people in the Justice Department. That year, Microsoft announced that it would buy Intuit, the maker of the financial program Quicken and one of the few independent suppliers of an application that had a commanding market share. After Microsoft initiated the purchase (and, critics charged, after learning the techniques of Intuit’s best programmers), the acquisition was dropped when the Department of Justice objected. Clearly sentiment was building up against Microsoft.

1997 年秋天，Internet Explorer 4.0 版问世。对于 Netscape 来说，情况比较复杂。捆绑 Internet Explorer 意味着微软犯了搭售罪，这使得 Netscape 无法销售其浏览器。但 1994 年 12 月，Netscape 在 Web 服务器上发布了 Navigator 的初始版本，用户可以免费下载。20^事实上，行业媒体称这是 Netscape 出色的营销创新。通过免费提供浏览器，Netscape 可以“锁定”客户，从那一刻起，这些客户就只能遵守 Netscape 标准。该浏览器对个人免费。企业只需支付少量许可费。该公司认为，一旦其浏览器成为标准，每个人都会为遵循该标准的其他 Netscape 产品付费。

That brings us to the fall of 1997 and Internet Explorer, version 4.0. For Netscape the case was complicated. Bundling Internet Explorer implied that Microsoft was guilty of a tie-in, making it impossible for Netscape to sell its browser. But when in December 1994 Netscape posted a preliminary version of its Navigator on a Web server, users could download it for free.²⁰ In fact, the trade press called that a brilliant marketing innovation by Netscape. By giving away the browser, Netscape would “lock in” customers who, from that moment onward, would be captive to a Netscape standard. The browser was free to individuals. Businesses were charged a modest licensing fee. The company assumed that once it established its browser as a standard, everyone would pay for other Netscape products that adhered to it.

一段时间内，这一策略非常有效。人们对网络浏览器，尤其是 Netscape 的兴趣如此之大，以至于该公司能够在 1995 年 8 月向公众发售股票，当时该公司还没有盈利。股价飙升，让其员工成为千万富翁（至少在纸面上）。互联网狂热开始了。

For a while, the strategy worked brilliantly. So great was the interest in Web browsers in general, and in Netscape in particular, that it was able to offer shares to the public in August 1995 before the company was profitable. The soaring price for the stock made multimillionaires of its employees (on paper at least). The Internet madness began.

微软专注于推出 Windows 95，但它也意识到了互联网正在发生的变化。微软的批评者们幸灾乐祸地指出，在1995 年出版的《未来之路》一书中，盖茨错过了这条道路上最重要的东西，即万维网。21^批评者们的言论是错误的：这本书经常描述基于联网计算机的未来，即使它在网络的细节上犯了错误。大多数批评者也没有注意到《未来之路》中盖茨写的 IBM 和数字设备公司未能察觉计算领域的巨变并因此遭受重创的段落。22^盖茨暗示微软面临着同样的风险。

Microsoft was focused on the introduction of Windows 95, but it was aware of what was happening to the Internet. Microsoft’s critics were gloating over how, in his book The Road Ahead published in 1995, Gates missed the biggest thing on that road, namely, the World Wide Web.²¹ The critics were off the mark: the book frequently describes a future based on networked computers, even if it got the details about the Web wrong. Most critics also failed to note the passages in The Road Ahead where Gates wrote of how IBM and Digital Equipment Corporation failed to sense a sea change in computing and stumbled badly.²² Gates implied that Microsoft faced the same risk.

随着 Internet Explorer 4.0 的发布，该公司的公关部门开始全力宣传盖茨如何与 DEC 的“英雄”肯·奥尔森不同，倾听战场上部队的呼声。23 故事讲述^了微软招聘人员如何访问大学校园，发现学生和教授通过电子邮件和文件传输完成课程。向媒体发布的内部备忘录和演讲记录显示，微软员工担心设计合理的 Web 浏览器可能会取代用户打开计算机时看到的 Windows 桌面。1995 年 12 月 7 日和 8 日，盖茨与分析师交谈，他宣称“沉睡的巨人已经醒来”——这是对 54 年前美国对日本袭击珍珠港的反应。24 ¹⁹⁹⁶年 3 月，微软赢得了一场关键战役，当时美国在线同意通过 Internet Explorer 而不是 Netscape 浏览器为其客户提供 Web 浏览服务。

As the release of Internet Explorer 4.0 approached, the company’s public relations apparatus kicked into high gear and touted the story of how Gates, unlike his “hero” Ken Olsen at DEC, listened to the message of his troops in the field.²³ The stories told of how Microsoft recruiters visited college campuses and found students and professors conducting their coursework by e-mail and file transfers. Internal memos and transcripts of speeches, released to the press, revealed a fear among Microsoft employees that a properly designed Web browser could replace the Windows desktop that users saw when they turned their computers on. On December 7 and 8, 1995, Gates spoke with analysts, to whom he declared “the sleeping giant has awakened”—a reference to the American response to the Japanese attack on Pearl Harbor fifty-four years before.²⁴ Microsoft won a key battle in March 1996, when America Online agreed to provide its customers with Web browsing through Internet Explorer instead of Netscape’s browser.

从此时起，事件的顺序变得模糊不清，不可能用几页纸来总结法院花了几年时间才理清的事情，尽管没有成功。关于审判的书籍已经很多，一旦审判结束，还会有更多。对微软的指控最简单的描述是 Internet Explorer 4.0 违反了《同意令》，因为它与 Windows 捆绑得太紧密。也就是说，到 1997 年，绝大多数使用 Windows 的个人电脑用户无法使用 Netscape 或任何其他浏览器轻松访问网络，而是被过度引导到 IE。指控的必然结果是，通过将 IE 捆绑到 Windows 中，该浏览器对个人和商业客户来说基本上是免费的——微软采取这一行动主要是为了切断 Netscape 的主要收入来源。

From this point on the sequence of events gets murky, and it is impossible to summarize in a few pages what it has taken the courts several years, unsuccessfully, to straighten out. A shelf of books has already appeared on the trial, and once it is settled there will be more. The simplest description of the charge against Microsoft is that Internet Explorer 4.0 violated the Consent Decree by being tied too closely to Windows. That is, personal computer users, who by 1997 were overwhelmingly using Windows, could not easily access the Web using Netscape’s or any other browser, but rather were steered too strongly to IE. A corollary to the charge was that by bundling IE into Windows, the browser was essentially free, for individuals as well as for business customers—an action that Microsoft took primarily to cut off Netscape’s major source of revenue.

在辩护中，微软犯下的错误导致检察官重新提起原本可能被掩盖的指控——例如，指控微软向第三方开发商隐瞒 API 细节。微软在开发操作系统 OS/2 期间与 IBM 的紧张关系也再次浮出水面。1997 年秋天，史蒂夫·鲍尔默在一次对员工的演讲中脱口而出，“去他妈的 [司法部长] 珍妮特·雷诺！”1998 年夏天，盖茨在视频中作证，在证词中，他显得紧张、闪烁其词、口齿不清——这与他精心打造的自信的公众形象截然相反。审判的很大一部分内容是讨论是否可以从 Windows 桌面上删除 IE 图标，以及如果可以删除，是否意味着微软正在遵守反对搭售的同意令。²⁵大量来自微软内部服务器的打印电子邮件信息被纳入记录，造成了进一步的尴尬。

In its defense, Microsoft made blunders that led prosecutors to resurrect charges that otherwise might have remained buried—charges over holding back the details of APIs from third-party developers, for example. Microsoft’s stormy relations with IBM during the development of the operating system OS/2 also resurfaced. In the fall of 1997 Steve Ballmer blurted out, in a speech to employees, “to heck with [Attorney General] Janet Reno!” In the summer of 1998, Gates gave a deposition on video, in which he appeared nervous, evasive, and inarticulate—the polar opposite of the confident public image he so carefully cultivated. A good portion of the trial was devoted to the question of whether one could remove the IE icon from the Windows desktop, and whether such a removal, if it could be done, implied that Microsoft was following the Consent Decree against a tie-in.²⁵ Reams of printed e-mail messages from Microsoft’s internal servers were introduced into the record, causing further embarrassment.

检方也犯了错误。最糟糕的是法官托马斯·潘菲尔德·杰克逊的一次采访，他在采访中直言不讳地表达了对微软的偏见。这足以让他在 2001 年 6 月推翻大部分判决，并将杰克逊从案件中剔除。

The prosecution made blunders, too. The worst was an interview by Judge Thomas Penfield Jackson, in which he flatly stated his prejudice against Microsoft. That was enough to get most of his judgment overturned in June 2001, and to have Jackson removed from the case.

无论法院的判决是什么，历史学家都没有义务将其视为具有约束力的历史判决。20 世纪 70 年代，法院裁定 John V. Atanasoff 而非 J. Presper Eckert 和 John Mauchly 是电子数字计算机的发明者。该判决具有重大的法律意义，但在历史学家中却得不到多少支持。如果法院在本案中判决微软败诉，历史学家可能会接受或不接受该判决，这取决于他们如何将其置于历史背景中。如果判决微软败诉，而该判决没有考虑到过去五十年计算机行业的创新速度，我会持怀疑态度。我持怀疑态度的原因之一是杰克逊法官在 1999 年 11 月发表的声明：“[Windows] 不存在任何商业上可行的替代品，[客户] 可以因微软大幅持续的价格上涨或类似措施而转而选择该替代品。” ²⁶这听起来太像 20 世纪 80 年代政府撤销对 IBM 的诉讼后，一位政府经济学家的声明了（第 8 章）。也许这次会是真的，但如果是这样，那将是计算机历史上的第一次。

Whatever the judgment of the court is, historians are not obligated to accept it as a binding judgment of history. In the 1970s, a court ruled that John V. Atanasoff, not J. Presper Eckert and John Mauchly, was the inventor of the electronic digital computer. That judgment had significant legal implications, but among historians it received little support. If the courts decide against Microsoft in the present case, historians may or may not accept that judgment depending on how they place it in historical context. I would be skeptical of a judgment against Microsoft that does not recognize the pace of innovation in the computer industry in the past fifty years. Among the reasons for my skepticism is a statement by Judge Jackson, made in November 1999, that “there exists no commercially viable alternative [to Windows] to which [customers] could switch in response to a substantial and sustained price increase or its equivalent by Microsoft.”²⁶ That sounds too close to the statement made by a government economist after the government dropped its suit against IBM in the 1980s (chapter 8). Perhaps it will be true this time, but if so it would represent a first for the history of computing.

从过去五十年的计算发展史来看，我们可以得出这样的结论：微软勇敢地试图避免 DEC 和 IBM 曾经陷入的陷阱，这可能给了它一些喘息的空间，但不会持续太久。它别无选择，只能适应互联网，并通过万维网让消费者能够访问它。我不会进一步猜测，但我想举一个例子来说明正在发生的事情。

From the perspective of the past fifty years of computing, one could conclude that Microsoft’s brave attempt to avoid the pitfalls that caught DEC and IBM may give it some breathing space but not for long. It has no choice but to accommodate itself to the Internet and its accessibility to the consumer via the World Wide Web. I will refrain from further speculation, but I do wish to mention one example that illustrates what is happening.

Hotmail、UNIX

Hotmail, UNIX

微软意识到了网络的威胁，于 1995 年推出了专有网络 MSN，并于 1996 年推出了“群件”通信系统 Exchange。这两个系统分别针对美国在线和 Lotus Notes。微软还推出了功能齐全的电子邮件程序 Outlook。但随着网络的爆炸式增长，微软不得不面对新的威胁：像 Yahoo! 这样的免费服务的出现，这些服务提供邮件、新闻、聊天和信息高速公路的友好入口（称为“门户”）。1997 年，微软以 4 亿美元收购了一个名为“Hotmail”的邮件程序，以应对这一威胁。Hotmail 已经迅速发展，并很快成为微软在网络上最大的业务。27 ^MSN被重新配置为互联网门户，而不是专有网络，微软在这些服务的市场份额开始迅速增长。Hotmail 不仅是免费的，而且与 Windows 的耦合度很低。它运行在 UNIX 机器上，而不是 Windows NT，微软收购它之后仍然如此。因此，甚至微软也违背了“吃自己的狗粮”这一神圣的格言。28^换句话说，盖茨夸口说他喜欢雇佣他能找到的最聪明的人，这很可能是真的，即使这意味着那些被雇佣的人会威胁到他公司的基础。

Recognizing the threat of networking, Microsoft introduced a proprietary network, MSN, in 1995, and a “groupware” communications system, Exchange, in 1996. These were aimed at America Online and Lotus Notes, respectively. It also introduced a full-featured e-mail program called Outlook. But as the Web exploded, Microsoft had to face a new threat: the advent of free services like Yahoo! that offered mail, news, chat, and a friendly on-ramp (called a “portal”) to the Information Highway. In 1997 Microsoft purchased (for $400 million) a mail program called “Hotmail” to counter this threat. Hotmail was already growing rapidly and soon became Microsoft’s biggest presence on the Web.²⁷ MSN was reconfigured to be an Internet portal rather than a proprietary network, and Microsoft’s market share for these services began to grow rapidly. Not only was Hotmail free, it was only loosely coupled to Windows. And it ran on UNIX machines, not Windows NT, and remained so after Microsoft bought it. Thus even Microsoft violated the sacred dictum of “eating your own dog food.”²⁸ In other words, Gates’s boast that he likes to hire the smartest people he can find is probably true, even if it means those hired will threaten the basis of his company.

这与 IBM 推出个人电脑时的情况相似，IBM 采用了英特尔处理器、微软软件、ASCII 代码和开放式架构。如果微软的收入来自 Windows 的连续版本和与之紧密结合的少数应用程序，那么它就必须适应，否则就会失去其作为领先个人电脑软件公司的地位。即使微软成功适应，它也将是一家不同的公司。再来看看 IBM 的例子，在 Louis Gerstner 的领导下，IBM 在 1991 年后成功适应了不断变化的计算领域。IBM 如今是一家成功且盈利的公司，但它已不再是 1980 年的大型机公司，而且它不再像批评者在针对它的诉讼被撤销后所指控的那样主宰和控制计算机行业^。29

The parallels with IBM’s introduction of the personal computer, with its Intel processor, Microsoft software, ASCII codes, and an open architecture should be obvious. To the extent that Microsoft derives its revenues from successive versions of Windows and a few applications tightly coupled to it, it will have to adapt or lose its place as the leading personal computer software company. Even if Microsoft does adapt successfully, it will be a different company. Again, consider the example of IBM, which under the leadership of Louis Gerstner successfully adapted to the changing computing field after 1991. IBM today is a successful and profitable company, but it is not the mainframe company it was in 1980, and it no longer dominates and controls the computer industry as its critics charged it would after the lawsuit against it was dropped.²⁹

点

Dot.Com

“我想我们中没有人知道这件事会如何发展，......但确实有令人兴奋的事情正在发生，而且是一件大事。”

—William Wulf，1993 年 5 月³⁰

“I don’t think any of us know where this thing is going anymore, ... but there’s something exciting happening, and it’s big.”

—William Wulf, May 1993³⁰

沃尔夫教授可能以为自己在夸大其词。其实不然。自从多萝西说她和托托已经不在堪萨斯之后，还没有人用如此轻描淡写的方式描述过如此重大的变化。

Professor Wulf probably thought he was exaggerating. He wasn’t. Not since Dorothy remarked that she and Toto were not in Kansas anymore has such a momentous change been described with such understatement.

互联网曾经只是学术界的少数教授或计算机行业的工程师们才知道的事情。我们中的许多人都还记得我们意识到互联网将成为一个更大的东西的那一天。那是在 1997 年 11 月的一个晚上，当时我刚完成本书第一版的手稿。我乘坐一辆配有司机的豪华轿车，准备向一群行业高管发表演讲。演讲的主题是计算机的历史，以及历史研究可以为规划未来提供哪些见解（如果有的话）。我准备了一些关于互联网历史的演讲，以及它将如何促进科学家、人文学者以及美国其他知识和专业精英之间的合作。在去酒店的路上，豪华轿车经过了一块灯火通明的广告牌，上面贴着一张巨大的沙奎尔·奥尼尔的照片，我隐约知道他是一名篮球运动员。广告牌的部分区域用胶合板条加长，以适应他向四面八方伸出的瘦长手臂和双腿。广告文字只有一句话：“ www.Shaq.com。” ³¹

The Internet was once something that a few professors in academia or engineers in the computer industry knew about. Many of us can remember the day when we realized it was going to be something bigger. That happened to me on an evening in November 1997, shortly after completing the manuscript for the first edition of this book. I was riding in a chauffeured limousine, on my way to speak before a group of high-level industry executives. The topic was the history of computing, and what insights, if any, the study of history could offer to chart the future. I had prepared some remarks about the history of the Internet, and about how it would facilitate collaboration among scientists, humanists, and others among the intellectual and professional elite in the country. On the way to the hotel the limo passed by a brightly lit billboard, on which was plastered a huge image of Shaquille O’Neal, who, I vaguely knew, was a basketball player. Parts of the billboard were extended with strips of plywood to accommodate his gangly arms and legs sticking out in all directions. The text of the advertisement consisted of one phrase: “www.Shaq.com.”³¹

等我回到酒店时，我意识到我的演讲已经过时了。直到那天晚上，我都是在一个狭隘的历史背景下理解互联网的：尝试远程访问计算机、建立防空和航空预订网络、分时共享主机以及共享昂贵的资源。现在互联网已经是另一回事了。它不再只是计算技术的一个方面；现在它已经成为娱乐、消费者支出和流行文化的一部分。互联网将计算与美国社会生活的主流融合在一起。

By the time I got to the hotel I realized my talk was obsolete. Until that night I had understood the Internet in a narrow historical context: of attempts to access computers remotely, to build air-defense and airline-reservation networks, to time-share mainframes, and to share expensive resources. Now the Internet was something else. It was no longer only a facet of computing technology; now it was part of entertainment, consumer spending, and popular culture. The Internet had fused computing with the mainstream of social life in America.

没有哪一件事，甚至沙奎尔·奥尼尔决定建立个人网页，能够将这一创新从一个方向转向另一个方向。事后看来，人们很容易说互联网的商业化是不可避免的，正如人们在回顾当时发生的令人困惑的事实时经常做的那样。事实上，如果不跨越一系列社会、政治和技术障碍，这种转变就不可能发生。互联网如此轻松地跨越了技术障碍，以至于人们往往甚至没有意识到这一点：它能够从处理由 56 Kbps 线路连接的几千个节点发展到由速度越来越快的卫星、微波和光纤线路连接的数百万个节点。很难找到另一种扩展性如此好的技术。互联网的扩展性源于其强大的设计，它将大多数网络活动放在与其相连的计算机和路由器上，而不是物理网络本身上。由于这些终端设备的能力和速度也随之增长（遵循摩尔定律），网络在 1969 年至 1996 年间的速度增长了 1000 倍，主机数量增长了 100 万个，且没有出现任何严重中断。32 1996 年以后的^持续增长，加上商业用途的增加，给网络带来了难以置信的压力，尽管它并不总是顺利运行，但仍在继续运行。

No single event, not even Shaquille O’Neal’s decision to mount a personal Web page, turned this innovation from one direction to another. In hindsight one can easily say that the commercialization of the Internet was inevitable, as people often do when looking back on the confusing tangle of facts as they happened. In fact such a transformation could not have occurred without jumping over a number of hurdles, social, political, and technical. The Internet jumped over the technical hurdle so easily that it is often not even acknowledged: its ability to evolve from handling a few thousand nodes linked by 56 Kbps lines to millions of nodes linked by ever-faster satellite, microwave, and fiber-optic lines. It would be hard to find another technology that scaled so well. The Internet scaled because of its robust design, one that put most of the network activities not on the physical network itself but on the computers and routers that were connected to it. Because these end devices, in turn, grew in capability and speed (following Moore’s law), the network was able to grow by a factor of 1,000 in speed and one million in number of hosts from 1969 to 1996, without experiencing any severe disruptions.³² Continued growth after 1996, plus increasing commercial use, have put incredible strains on the network, yet it continues to function, although not always smoothly.

可接受使用政策

The Acceptable Use Policy

政治障碍在于如何在一个由联邦政府签订合同构想和建设的网络上容纳明显的商业流量。20 世纪 80 年代初，网络的重点从 ARPA 转移到了美国国家科学基金会，后者从 1988 年到 1995 年管理着一个名为 NSFnet 的网络。NSF 于 1990 年承担了互联网的责任，原来的 ARPANET 被退役（军方开发了自己的受限网络）。NSF 必须解决如何处理商业公司连接和使用互联网的问题。它以“可接受使用政策”作为回应，其中部分内容如下：“NSF 主干服务用于支持美国研究和教学机构内部和之间的开放研究和教育，以及营利性公司的研究部门在从事开放学术交流和研究时。不得将其用于其他目的。” ³³

The political hurdle was how to accommodate obvious commercial traffic on a network that was conceived and built by contracts let by the federal government. In the early 1980s the focus of networking shifted from ARPA to the National Science Foundation, which managed a network called NSFnet from 1988 through 1995. The NSF assumed responsibility for the Internet in 1990, and the original ARPANET was decommissioned (the military evolved its own, restricted networks). The NSF had to address the question of how to deal with commercial firms being connected to and using the Net. It responded with an “Acceptable Use Policy,” which read in part: “NSF Backbone services are provided to support open research and education in and among U.S. research and instructional institutions, plus research arms of for-profit firms when engaged in open scholarly communication and research. Use for other purposes is not acceptable.”³³

该政策允许“发布新产品或活动公告……但不允许发布任何形式的广告”。因此，IBM 可以使用互联网传播其某款产品的技术信息，尤其是如果此举可以帮助用户将该产品连接到互联网。它可以宣布新 PC 上市，但不能宣布年底促销。这条界限并不明确，但 NSF 还是试图划定了界限。该政策进一步允许“除非法或明确不可接受的使用外，允许进行其他可接受用途的偶然通信”。这意味着允许使用个人电子邮件甚至讨论组，只要它们不占据特定网站的流量。“大量用于私人或个人业务”被明确视为不可接受。奥尼尔不得不等待。

The policy allowed “announcements of new products or activities ... but not advertising of any kind.” Thus it was all right for, say, IBM to use the Internet to disseminate technical information about one of its products, especially if that would help users connect that product to the Internet. It could announce the availability of new PCs but not announce a year-end sale on them. The line was not clear, but the NSF tried to draw it anyway. The policy further allowed “communication incidental to otherwise acceptable use, except for illegal or specifically unacceptable use.” That implied that personal e-mail and even discussion groups were allowed, as long as they did not dominate the traffic to or from a particular site. “Extensive use for private or personal business” was specifically deemed unacceptable. Shaq would have to wait.

到 1992 年，这些限制被取消。互联网上的流量本来就增长迅速，现在增长得更快——从 1992 年 1 月的每月 1 万亿次增加到 1994 年的每月 10 万亿次。本节开头引用的 Wulf 教授曾是 DEC 的工程师，当时他正在弗吉尼亚大学的学术岗位上休假，并在 20 世纪 80 年代末负责 NSF 的网络项目。和这个以研究为导向的联邦机构的其他人一样，他怀着既害怕又兴奋的心情看待 NSFnet 上流量的增长。自 17 世纪以来，科学知识总体上呈指数级增长，但速度却没有达到这样的水平。NSF 必须赶在列车加速之前下车。1995 年，NSFnet 解散，NSF 退出了网络运营业务，重新开始资助研究。互联网被私有化了。

By 1992 the restrictions were lifted. Traffic on the Internet, already growing rapidly, grew even faster—from one trillion byes a month in January 1992 to ten trillion a month in 1994. Professor Wulf, quoted at the beginning of this section, was a former DEC engineer, on leave from an academic post at the University of Virginia, and in charge of the NSF’s networking program in the late 1980s. Like the others at the research-oriented federal agency, he looked at the growth of traffic on NSFnet with a mixture of fear and excitement. Scientific knowledge in general has been growing exponentially since the seventeeth century, but not at these rates. The NSF had to get off the train before it accelerated any faster. In 1995 the NSFnet was dissolved, and the NSF got out of the business of running a network and back to funding research. The Internet was privatized.

但是，如何实现的呢？这一权力转移的细节尚不明确。部分混乱来自副总统艾尔·戈尔（Al Gore, Jr.）在 1999 年 3 月接受 CNN 的沃尔夫·布利策（Wolf Blitzer）采访时的说法，人们认为戈尔试图声称对这一权力转移负责。戈尔并没有像批评者所指责的那样声称他“发明了互联网”，但他给人的印象就是这样，媒体也报道了这一点。确切的说法是：“在我任职于美国国会期间，我率先创建了互联网。”戈尔的失误无助于他的竞选。34^作为一名经验丰富的政治家，他知道媒体可以如何歪曲报道，但他显然不明白的是，公众对任何重大技术的“发明”所带来的细微差别几乎一无所知，也难以容忍。

But how? The particulars of this transfer are murky. Some of the confusion comes from a claim made by Vice President Al Gore, Jr., who people thought tried to claim responsibility for this transition, in an interview with Wolf Blitzer of CNN in March 1999. Gore did not claim that he “invented the Internet,” as his critics charged, but that was the impression he gave and that was how the press reported it. The exact words were, “During my service in the United States Congress, I took the initiative in creating the Internet.” Gore’s blunder did not help his candidacy.³⁴ As a seasoned politician he knew how the press could distort a story, but what he apparently did not understand was that the public has little understanding of—or tolerance for—the nuances that accompany the “invention” of any major technology.

回顾整个骚乱，戈尔似乎想将简化向公共用途过渡的功劳归于自己：这一过渡是随后的“革命”所依赖的，显然需要某种立法行动才能实现。在 1998 年为教育电视制作的一部关于互联网的电视连续剧中，斯蒂芬·西加勒声称，关键立法不是来自戈尔，而是来自弗吉尼亚州众议员里奇·鲍彻，后者于 1992 年 6 月提出了一项立法修正案，授权 NSF“支持计算机网络的开发和使用，这些网络可能主要用于科学和工程研究和教育以外的目的。” ³⁵据西加勒称，当乔治·H·W·布什总统签署该法案成为法律时，它实际上终止了《可接受使用政策》。这可能是实现这一变化的法律，但戈尔，而不是鲍彻，发挥了更重要的作用。

Looking back on the whole brouhaha, it appears that Gore was trying to claim credit for easing the transition to public usage: a transition on which the subsequent “revolution” depended, and one that obviously required some sort of legislative action to effect. For a television series on the Internet, produced in 1998 for educational television, Stephen Segaller claimed that the crucial legislation came not from Gore but from Congressman Rich Boucher of Virginia, who in June 1992 introduced an amendment to legislation that authorized the NSF to “support the development and use of computer networks which may be used substantially for purposes in addition to research and education in the sciences and engineering.”³⁵ According to Segaller, when President George H. W. Bush signed the bill into law, it effectively ended the Acceptable Use Policy. That may have been the law that effected the change, but Gore, not Boucher, played a more important role.

甚至连戈尔的批评者也承认，在担任副总统之前，戈尔担任参议员时，是联邦政府支持计算机网络的坚定支持者。尽管他没有创造“信息高速公路”一词，但他孜孜不倦地推广这一概念，并负责让该词成为通用词汇。36^他对媒体失言的一个奇怪之处在于，没有人报道戈尔以及美国国家科学基金会和其他地方的许多其他人设想的未来互联网几乎与现实完全相反。简要总结一下这一系列复杂且快速发展的事件，戈尔的愿景让人想起了 ARPANET 的早期。他希望联邦政府协助建设一个高速网络，称为国家研究和教育网络 (NREN)，这将使研究人员能够访问稀缺资源，尤其是昂贵的超级计算机。37^有了这个网络，全国各地的科学家就可以推动物理学、化学以及最重要的生物医学研究的前沿发展。反过来，NSF 将退出运营和支付网络费用的业务，但会坚持要求科学家自己支付他们所需的任何网络费用。当然，他们可以将这些费用作为向 NSF 或任何其他资助机构提交的拨款申请的一部分，人们认为电信公司会制定一个物理计划来响应这个市场。不仅发生了这种情况，而且随着互联网向商业流量开放，人们争相建造此类设施。（太多公司加入进来，导致泡沫在 2001 年破灭。）最终，对超级计算机或其他稀缺资源的需求与许多应用程序对 PC 和工作站的一般互联网访问的需求相比很小，其中科学研究是少数。向科学开放网络的影响是巨大的；与互联网向公众开放时发生的其他事情相比，它似乎微不足道。

Even Gore’s critics admit that as a senator, before he became vice president, he was a fierce champion of federal support of computer networking. If he did not coin the phrase “Information Superhighway,” he promoted the concept tirelessly and was responsible for bringing that phrase into common currency.³⁶ One curious aspect of his gaffe to the press was that no one reported that Gore, along with many others at the NSF and elsewhere, envisioned a future Internet that was nearly the opposite of how things turned out. To summarize briefly the complex and rapidly evolving series of events, Gore’s vision was reminiscent of the earliest days of the ARPANET. He wanted the federal government to assist in building a high-speed network, called the National Research and Education Network (NREN), which would allow researchers to gain access to scarce resources, especially expensive supercomputers.³⁷ With that net in place, scientist all across the country could push the frontiers of physics, chemistry, and above all biomedical research. The NSF in turn would get out of the business of running and paying for a network but would insist that the scientists themselves pay for whatever networking they needed. They could, of course, include those charges as part of the grant applications to the NSF or any other funding agency, and people assumed that telecommunications companies would build a physical plan to respond to this market. Not only did that happen, but with the opening up of the Internet to commercial traffic, there was a land rush to build such facilities. (Too many companies jumped in, and the bubble burst in 2001.) Ultimately, the demand for access to supercomputers or other scarce resources was small compared to the demand for general Internet access on PCs and workstations for many applications, of which scientific research was in the minority. The impact of opening up networking to science was enormous; it only seems small in comparison to the other things that happened when the Internet was opened to public access.

1991 年，担任参议员的戈尔提议建立国家信息基础设施，以使这一转变正式化。高性能计算法案（“戈尔法案”）的主要部分经过 1992 年的辩论，最终通过了一个版本。与此同时，戈尔离开参议院，于 1993 年 1 月出任副总统。38^作为副总统，他继续倡导使用互联网，坚持要求联邦机构建立网页，包含有关它们是谁、它们做什么的基本公共信息。白宫在www.whitehouse.govi建立了一个网页，其中展示了第一家庭的猫（Socks）的照片等。当有人点击这只猫的图片时，它会喵喵叫。这在 21 世纪听起来不算什么，但在 20 世纪 90 年代初的个人计算背景下，这是一个重大进步。

While a senator in 1991, Gore proposed a bill to create what he called a National Information Infrastructure, which would formalize this transition. The essential parts of the High Performance Computing Act (the “Gore bill”) were debated through 1992, and a version was eventually passed. Meanwhile, Gore left the Senate and became vice president in January 1993.³⁸ As vice president he continued to champion Internet usage, insisting that federal agencies set up Web pages containing basic public information about who they were and what they did. The White House set up a Web hpage at www.whitehouse.govi, which among other things, showed a picture of the First Family’s cat (Socks). When someone clicked on the cat’s image, it meowed. That does not sound like much in the twenty-first century, but in the context of personal computing in the early 1990s it was a major advance.

亚历山大·格雷厄姆·贝尔 (Alexander Graham Bell) 认为电话主要是一种商业工具，他惊讶地发现人们用它来闲聊。托马斯·爱迪生 (Thomas Edison) 并没有想到他的留声机会用于音乐和娱乐。同样，互联网的发明者也没有预见到万维网的商业用途（互联网有很多“发明者”，当然没有像爱迪生或贝尔这样的个人）。用埃德·坦纳 (Ed Tenner) 的话来说，这些“未预料到的后果”的象征发生在网络冲浪者试图访问hwww.whitehouse.comi而不是hwww.whitehouse.govi时：他们被带到了一个提供色情材料（收费）的网站。色情推动了互联网的大部分商业化，就像它推动了视频录制和电影的早期一样。³⁹ 1992 年，注册的 .com 网站数量远远落后于 .edu（尽管领先于 .gov），但到了 20 世纪中叶，.com 网站已经压倒了所有其他网站，因此人们在试图联系白宫时输入这个后缀也就不足为奇了^。40

Alexander Graham Bell thought the telephone would primarily be a business tool and was surprised to find people using it for idle chat. Thomas Edison did not envision his phonograph being used for music and entertainment. Likewise, the commercial use of the World Wide Web was not foreseen by the Internet’s inventors (and it had many “inventors,” certainly not a single individual like an Edison or Bell). Symbolic of these “unanticipated consequences,” to use Ed Tenner’s phrase, occurred when Web surfers tried to go to hwww.whitehouse.comi instead of hwww.whitehouse.govi: they were taken to a site offering pornographic materials (for a fee). Pornography drove much of the commercialization of the Internet, just as it did the early days of video recording and of motion pictures.³⁹ In 1992, the number of registered dot.com sites was well behind dot.edu (although ahead of dot.gov), but by mid-decade the dot.com world overwhelmed all the others, so it was no surprise that people typed in that suffix when trying to reach the White House.⁴⁰

Java

前面关于商业化如何进入互联网的讨论并没有考虑到“互联网”和公众对互联网的认知（通过万维网访问）之间的区别。两者是不同的，应该理解这种区别。商业活动几乎全部通过网络进行。Tim BernersLee 的超文本传输协议 (HTTP) 压倒了使用 FTP、Gopher、WAIS 或远程登录的互联网交易。然而，并非只有网络的发明才为商业使用铺平了道路。在 20 世纪 90 年代初，正当这种现象开始出现时，SUN Microsystems 的 Bill Joy 意识到需要一种符合时代的编程语言。他和其他人一样不确定互联网的发展方向，但他认为现有的语言无法胜任这项任务。他说需要一种语言，既能保留当时迅速流行的 Cþþ 的先进性，又能拥有更多 C 或汇编语言的底层功能。他号召程序员编写一种语言，他称这种语言是“C 语言++-”。⁴¹从 1991 年开始，James Gosling 和 SUN 的一个小团队开发了一种名为 Oak 的语言，满足了 Joy 的需求。在 Joy 的支持下，该语言经过重新设计，更名为“Java”（出于法律原因），并于 1995 年 3 月公开发布。就在互联网允许商业使用、万维网使导航变得容易的那一刻，Java 应运而生：这种语言使 Web 设计师能够在其产品中加入“亮点”。众所周知，卖点是“亮点”，而不是“牛排”。

The preceding discussion of how commercialism came to the Internet does not respect the distinction between “the Internet” and the public perception of the Internet, accessed through the World Wide Web. The two are different and that distinction should be understood. Commercial activities, almost exclusively, are done via the Web. Tim BernersLee’s Hypertext Transfer Protocol (HTTP) overwhelms Internet transactions that use FTP, Gopher, WAIS, or remote login. It was not just the Web’s invention that paved the way for commercial use, however. In the early 1990s, just as this phenomenon was starting, Bill Joy of SUN Microsystems recognized a need for a programming language that would fit the times. He was no more certain of where the Internet was going than anyone else, but he believed that existing languages were not up to the task. He spoke of a need for a language that retained the advances of Cþþ, then rapidly gaining popularity, but with more of the low-level power of C or assembly language. He called on programmers to write a language that was, he said, “C-plus-plus-minus.”⁴¹ Beginning in 1991 James Gosling, along with a small team of other programmers at SUN, came up with a language, called Oak, that filled Joy’s needs. With Joy’s support, it was reworked, renamed “Java” (for legal reasons), and publicly announced in March 1995. At the precise moment that commercial uses were being allowed on the Internet, and as the World Wide Web made navigating easy, along came Java: a language that enabled Web designers to put the “sizzle” in their offerings. As everyone knows, it is the “sizzle”—not the “steak”—that sells.

Java 很快成为 Web 设计师赋予其页面动画、移动和交互性的一种手段。它之所以流行起来，是因为用 Java 编写的程序几乎可以在任何联网的计算机（无论大小）上运行，无论供应商是什么。与互联网本身的底层设计一样，Java 利用 PC 和工作站日益强大的功能来提供翻译功能，因此 Java 程序员只需“编写一次，随处运行”。它使用的概念至少可以追溯到个人计算的初期。回想一下，当 IBM PC 于 1981 年发布时，客户可以选择三种操作系统。除了微软的 DOS 和 Digital Research 的 CP/M，还可以获得 UCSD（加利福尼亚大学圣地亚哥分校）的“p 系统”。这个用 Pascal 编写的系统不是将代码翻译成机器语言，而是将其翻译成中间“伪”机器（因此得名）的代码，然后再编译和执行。为什么要增加这一层复杂性？原因是，通过让每个计算机制造商编写自己的 p 代码编译器，操作系统可以在任意数量的计算机上运行而无需修改。

Java quickly became the means by which Web designers could give their pages animation, movement, and interactivity. It caught on because a program written in Java could run on nearly any computer, large or small, from any vendor that was connected. As with the underlying design of the Internet itself, Java took advantage of the growing power of the PCs and workstations to provide the translating capabilities, so that the Java programmer cold simply “write it once, run it anywhere.” It did that by using a notion that went back at least to the beginnings of the personal computing. Recall that when the IBM PC was announced in 1981, customers had a choice of three operating systems. Besides Microsoft’s DOS and Digital Research’s CP/M, one could get the UCSD (University of California, San Diego) “p-system.” This system, written in Pascal, translated code not into machine language but into code for an intermediate “pseudo” machine (hence the name), which in turn was compiled and executed. Why this extra layer of complexity? The reason was that by letting each computer manufacturer write its own p-code compiler, the operating system would run on any number of computers without modification.

p 系统从未流行起来。原因之一是 IBM PC 很快就成为标准，而其他所有可能利用 p 系统的机器都没有站稳脚跟。另一个原因是 p 系统的两阶段翻译过程导致速度下降，与 MS-DOS 相比速度慢得令人无法接受。尽管 Pascal 语言在大学里备受推崇，但 PC 程序员并不关心它；他们更喜欢 MS-DOS 和 Microsoft 的 BASIC 的原始、接近金属的代码。

The p-system never caught on. One reason was that the IBM PC quickly became a standard, and all those other machines that might have taken advantage of the p-system never established themselves. Another reason was that the speed penalty, which the p-system’s two-stage translation process required, made it unacceptably slow compared to MS-DOS. And PC programmers did not care for the Pascal language, though it was admired in the universities; they preferred the raw, close-to-the-metal code of MS-DOS and Microsoft’s BASIC.

十二年后——在“互联网时代”中，这已经是永恒——情况发生了变化。网络现在承载着各种各样、各种尺寸的机器——甚至有计划为电视、手持式记事本和手机提供网络访问。硬件和软件供应商看到网络访问如何促进他们的销售，并不介意编写代码来从 Java 伪机器进行翻译。最后，奔腾处理器的强大功能足以克服双重翻译的低效率。请注意，与为特定机器量身定制的程序相比，低效率仍然存在，并且将永远存在。但考虑到 20 世纪 90 年代中期处理器能力、编译器和语言设计以及电信速度的结合，这次它不那么重要了。

A dozen years later—an eternity in “Internet time”—the situation had changed. The Web was now hosting machines of a great variety and size—there were even plans to provide Web access to televisions, handheld organizers, and cell phones. Suppliers of hardware and software, seeing how Web access boosted sales for them, did not mind writing the code to translate from the Java pseudomachine. Finally, the brute force of the Pentium processor was enough to overcome the inefficiencies of a dual translation. Note that the inefficiency is still there and will always be there, when compared to a program tailored for a specific machine. But given the mix of processor power, compiler and language design, and telecommunications speeds of the mid-1990s, it mattered less this time.

随着 Java 的流行，媒体对它的关注度远远超过了它的实际功能，这是很不幸的。Java 的“一次编写，随处运行”特性在行业媒体中被宣扬为一种在 Web 上做一些有趣的事情的方式，但它打破了微软对个人计算软件的控制。如果人们可以在大型服务器上编写程序，并通过互联网将这些程序发送到桌面，谁还需要从微软购买 Office 套件呢？如果人们可以用 Java 编写程序，并且任何计算机都可以运行，谁还需要 Windows？这与人们对 Netscape 的 Navigator 的评价不同，在这两种情况下，他们都错了。微软对 Java 的流行并不热衷，尽管他们获得了 SUN 的使用许可。SUN 后来声称微软违反了协议，这些论点与 Netscape 的论点一起进入了法庭。对于倒霉的网络冲浪者来说，Java 是一把双刃剑：它确实提供了“刺激”，但许多 Web 设计师用它来设计的网站几乎没有其他东西。在等待 Java 繁重的页面通过缓慢的电话连接加载时，冲浪者在 Al Gore 的信息高速公路上第一次体验到了拥堵。Java 尚未证明其在制造“智能”电器（如烤面包机、恒温器甚至手机）方面的价值，但那些想在互联网上销售商品的人立即接受了它。

As Java caught on, it garnered media interest all out of proportion to what the language actually did, and that was unfortunate. Java’s write-once, run-anywhere feature was heralded in the trade press not as a way to do something interesting on the Web, but to break Microsoft’s hold on personal computing software. If people could write programs on large servers, and have those programs sent to the desktop over the Internet, who needed to buy the Office suite from Microsoft? If people could write programs in Java, which any computer could run, who needed Windows? It was a variant of what people were saying about Netscape’s Navigator, and in both cases they were wrong. Microsoft was not enthusiastic about Java’s popularity, although they got a license from SUN to use it. SUN later claimed that Microsoft violated the agreements, and these arguments made their way into the courtroom along with those coming from Netscape. For the hapless Web surfer Java was a mixed blessing: it provided “sizzle,” all right, but a lot of Web designers used it to design sites that had little else. Waiting for a Java-heavy page to load through a slow telephone connection, surfers experienced the first evidence of gridlock on Al Gore’s Information Superhighway. Java has yet to prove its worth in making “smart” appliances like toasters, thermostats, or even cell phones, but people with things to sell on the Internet embraced it immediately.

所有这些条件都具备后，互联网泡沫自然而然地出现了。美国计算机行业已经经历了几次这样的泡沫。我们在前面的章节中已经记述了大部分泡沫：20 世纪 50 年代末的鼓式真空管计算机、20 世纪 60 年代“gogo”时期的小型计算机公司、20 世纪 70 年代末的 Altair 兼容 PC、20 世纪 80 年代中期的“JAWS”（只是另一种工作站）现象，以及同样在那个十年中出现的 IBM PC 个人软件。

With all these pieces now in place, the dot.com bubble followed naturally. The U.S. computer industry had already seen several of these. We have chronicled most in earlier chapters: drum-based vacuum-tube computers in the late 1950s, minicomputer companies during the “gogo” 1960s, Altair-compatible PCs in the late 1970s, the “JAWS” (just another workstation) phenomenon in the mid-1980s, and personal software for the IBM PC also in that decade.

1995 年 7 月，杰夫·贝佐斯开设了一家名为 Amazon.com 的在线书店。到 10 月份，每天处理 100 份订单。据该公司称，到 2000 年 7 月，“一分钟处理 100 份订单已很常见”。尽管贝佐斯于 1999 年 12 月被《时代》杂志评为“年度人物”，但该公司尚未实现盈利^。42 1995 年 9 月，皮埃尔·奥米迪亚 (Pierre Omidyar) 推出了一项名为“拍卖网”的在线拍卖服务，他希望通过这项服务帮助女友交易 Pez 糖果盒（他声称如此；批评者声称这个故事是捏造的）。拍卖网发展成为 eBay，2001 年进行了 700 万次拍卖，交易的物品从棒球卡到新车、二手车和老式汽车，再到湾流喷气式飞机（售价 490 万美元）。43^与亚马逊和许多其他商业网站不同，eBay 是盈利的。它向卖家收取少量费用以列出商品，如果商品售出，则根据销售价格再收取另一笔费用。亚马逊和 eBay 是少数几家拥有稳定收入来源的网络企业（色情网站除外）。大量依赖广告的网站（其广告通常由 Java 驱动）表现不佳；到 2001 年夏天，许多网站陷入财务困境或倒闭。在撰写本文时，互联网泡沫仍在继续，因此现在判断哪些初创公司将幸存下来以及哪些公司将在这场动荡中脱颖而出还为时过早。除了上面提到的 eBay 和亚马逊之外，互联网“门户”雅虎似乎已经实现了规模和稳定，尽管它和亚马逊一样没有盈利，而且前景不明朗。

In July 1995 Jeff Bezos opened an on-line bookstore he called Amazon.com. By October it was processing 100 orders a day. According to the company, by July 2000, “a 100-order minute [was] common.” The company has yet to demonstrate profitability, although Bezos was named Time Magazine’s “Person of the Year” in December 1999.⁴² In September 1995 Pierre Omidyar started an on-line auction service called “Auction Web” that he hoped would help his girlfriend trade Pez dispensers (or so he claimed; critics claim that this story was a fabrication). Auction Web grew into eBay, with seven million ongoing auctions in 2001, trading items from baseball cards to new, used, and vintage automobiles, to a Gulfstream jet (which sold for $4.9 million).⁴³ Unlike Amazon and many other commercial sites, eBay is profitable. It levies a small charge to the seller to list an item, and levies another charge based on the selling price if the item is sold. Amazon and eBay were among the few Web businesses (other than pornographers) with a steady revenue stream. The myriad of sites that relied on advertising—their ads often driven by Java—did not fare as well; many were in financial trouble or out of business by the summer of 2001. At the time of this writing the dot.com collapse is still in force, so it is too early to tell who among the start-up companies will survive and what will emerge from all the turmoil. Besides eBay and Amazon, mentioned above, the Internet “portal” Yahoo! seems to have achieved bulk and stability, although like Amazon it has been unprofitable and does not have a clear future.

如果说在幸存下来的商业网站中有什么共同点，那就是它们为用户提供了一种社区感。它们的核心仍然是商业网站，但用户至少可以对 ARPANET 以及在第一批 Usenet 网站上创建的在线社区有一点了解。除了 Usenet 和前面提到的在线公告板之外，还有一个社区脱颖而出，成为其先驱，是互联网版的 Homebrew 计算机俱乐部。那就是位于加利福尼亚州索萨利托的 WELL（“Whole Earth 'Lectronic Link”）。它成立于 1984 年底，是《全球概览》的电子版。再次，帮助创办并命名它的斯图尔特·布兰德在将计算机技术与反主流文化价值观融合方面发挥了关键作用。用户只需支付少量订阅费，它就可以在当时缓慢的拨号调制解调器连接上运行。从一开始到现在（它仍然活跃），它不仅以其湾区反主流文化风格而闻名，还因其帖子的质量而闻名。它吸引了大量作家，并且仍然专注于写作，而不是图形或多媒体。它是感恩至死乐队的一群特别活跃的粉丝的家园，他们交换有关该乐队的信息。根据那个时代的记载，来自 Deadheads 的流量使 WELL 在危险时期保持了财务偿付能力^。44

If there is any pattern to be found among the commercial Web sites that have survived, it is that they provide their patrons with a sense of community. They remain commercial sites at their core, but users get at least a vestigial sense of what ARPANET must have been like, of the online communities created on the first Usenet sites. Besides Usenet and the on-line bulletin boards already mentioned, one community stood out as a precursor, and was the Internet equivalent of the Homebrew Computer Club. That was the WELL (“Whole Earth ’Lectronic Link”), located in Sausalito, California. It was founded in late 1984 as an electronic version of the Whole Earth Catalog. Once again, Stewart Brand, who helped start it and gave it its name, played a key role in fusing computer technology with countercultural values. Users paid a modest subscription fee, and it ran over the slow dial-up modem connections of the day. From its beginnings, and on to the present day (it is still active), it was notable not just for its Bay-area counter-cultural flavor but also for the quality of the postings. It attracted a large number of writers and remains focused on writing, not graphics or multimedia. It was the home for an especially active group of fans of the Grateful Dead, who traded information about the band. According to chronicles of the era, the traffic from Deadheads kept the WELL financially solvent through perilous times.⁴⁴

WELL 的成员中有霍华德·莱茵戈尔德，他的作品提倡“虚拟社区”的概念。另一个成员是约翰·佩里·巴洛，他不仅是死者乐队的成员，甚至还为死者乐队写过歌词（当然，是在乐队后期的巴洛克时期）。巴洛与米奇·卡普尔共同创立了电子前沿基金会，1996 年，他在网络上发布了“网络空间独立宣言”，谴责那些试图扼杀蓬勃发展的互联网的商业和政府。4 ⁵宣言的开头是这样的：“工业世界的政府，你们这些令人厌倦的血肉和钢铁巨人，我来自网络空间，这是思想的新家园。代表未来，我请求你们这些过去的人离我们远点。你们不受欢迎。在我们聚集的地方，你们没有主权。”

Among the members of the WELL was Howard Rheingold, whose writings promoted the concept of “virtual communities.” Another was John Perry Barlow, who was not only a Deadhead but even wrote lyrics for the Dead (to be sure, in the band’s later, baroque period). Barlow was a cofounder, with Mitch Kapor, of the Electronic Frontier Foundation, and in 1996 he posted on the Web his “Declaration of the Independence of Cyberspace,” a rant against those in the commercial and government world who were trying to muzzle the burgeoning Internet.⁴⁵ It opened with these words: “Governments of the Industrial World, you weary giants of flesh and steel, I come from Cyberspace, the new home of the Mind. On behalf of the future, I ask you of the past to leave us alone. You are not welcome among us. You have no sovereignty where we gather.”

这一现象再次提醒我们，非技术因素在个人电脑发明时发挥了巨大作用，如今仍在影响计算的发展方向。20 世纪 70 年代，大型计算机公司的高管们无法理解 Apple II 为何能取得成功。20 世纪 90 年代，正是这些网络公司和网络本身创造了一种新的商业模式，而当时行业领袖们都在吹嘘 500 频道有线电视或“视频点播”之类的东西。

Once again this phenomenon should remind us that nontechnical factors, so strong at the invention of the personal computer, continue to shape the direction of computing. In the 1970s, executives at the big computer companies could not understand how the Apple II could ever be successful. In the 1990s it was those Web companies, and the Web itself, that created a new commercial paradigm, at a time when industry leaders were touting things like 500-channel cable television or “video on demand.”

在亚马逊网站上，客户被鼓励撰写书评。更巧妙的是，其他人可以对评论进行评分。一篇批评一本书的评论，反过来又被亚马逊“社区”的其他成员评为低分，因此可以得到相应的评判。包括前众议员纽特·金里奇在内的一些人因其评论的数量和质量而闻名。大学教授发现学生写的研究论文提到了他们既没有借过、买过也没有读过的书：他们只是看了亚马逊网站上的评论并总结了内容。46^患有写作障碍的作家找到了一个不写作的新借口：他们每天会多次查看亚马逊网站，看看他们现有的书卖得好不好。同样，eBay 设计了一个客户反馈系统，对拍卖品的可信度进行排名。一个迟交货物的卖家，或者虚假宣传收藏娃娃的状况的卖家，很快就会受到社区的谴责。结果是降低了欺诈率，而二手商品拍卖是欺诈行为普遍存在的领域。客户自己做这些工作，eBay 不太可能如此有效地监管其网站。其他网站让用户购买和评论消费品，然后让其他人对评论者的可信度进行评分。还有一些网站提供“专家”的建议，这些专家的声誉和权威并非来自他们的学位或职位，而是来自过去听取他们建议的人对他们的建议的评价。有多少这样的网站能够将这种社区感转化为盈利能力是一个悬而未决的问题，但没有提供社区感的网站很难维持业务。

On the Amazon site, customers were encouraged to write reviews of a book. Even more ingenious, others could rate a review. A review that panned a book, which in turn was rated low by other members of Amazon’s “community,” could thereby be judged accordingly. Some, including former Representative Newt Gingrich, became well known for the quantity and quality of their reviews. College professors found students writing research papers that referred to books they had neither borrowed, bought, nor read: they simply looked at the reviews on Amazon.com and summarized the contents.⁴⁶ Authors afflicted with writer’s block found a new excuse not to write: several times a day they would check Amazon’s site to see how well (or poorly) their existing books were selling. Likewise, eBay devised a system of customer feedback that ranked the credibility of those auctioning items. A seller who was late delivering goods, or who falsely advertised the condition of a collectible doll, would soon be chastised by the community. The result was to lower the rate of fraud, in a field—auctions of used goods—where fraud is common. The customers did the work, and it is unlikely that eBay could have been so effective in policing its site. Other Web sites let users buy and review consumer items, then let others rate the credibility of the reviewer. Still others provided advice by “experts,” whose reputation and authority came not from their academic degree or position but from how well their advice has been rated by those who followed it in the past. How many of these sites will be able to convert that communal feeling into profitability is an open question, but sites that do not offer a sense of community had trouble staying in business.

随着个人电脑进入企业的办公桌，它们也继承了创建首批个人电脑的黑客的精神。同样，网络的商业化也继承了第一批在线社区的社区精神。当然，这种精神是不一样的，一些参与创建网络的人并不乐意看到它成为一种大众现象。尽管如此，约翰·佩里·巴洛的《宣言》中有一小段关键的内容嵌入了商业化的网络中，对这种精神的理解，以及制定合理商业计划等更明显的因素，将迄今为止在这场大混乱中幸存下来的少数商业网站与未能幸存下来的网站区分开来。

As PCs found their way onto the desks of corporations, they carried with them the spirit of the hackers who created the first of their kind. So too did the Web’s commercialization carry the communal spirit of the first on-line communities. That spirit was not the same, of course, and some who had been present at the creation of were unhappy to see it become a mass phenomenon. Still, a small but critical piece of John Perry Barlow’s Declaration is embedded in the commercialized Web, and an understanding of this spirit, along with the more obvious factors of developing a sound business plan, and so on, separated the few commercial sites that have so far survived the shakeout from those who have not.

搜索引擎、门户网站

Search Engines, Portals

除了上述商业网站之外，另一类网站是搜索引擎和“门户网站”：帮助人们浏览网络并找到他们感兴趣的信息的网站。其中最成功的是雅虎！，由 Dave Filo 和 Jerry Yang 在斯坦福大学读书时于 1994 年创立，当时名为“杰里的万维网指南”。随着网络的发展，出现了许多搜索引擎。大多数搜索引擎使用某种自动化程序（“bot”，机器人的缩写）来“爬行”网站并提取关键字。雅虎也使用自动化程序，但其索引是由杨致远和 Filo 编制的——起初是亲自编制，后来由其他人补充。因此，从一开始，雅虎就不仅仅是一个搜索引擎，而是一个以人性化的方式消除了人们赤身裸体访问网络的恐惧的地方。所有这些都是免费提供的，并且一直如此。1995 年 8 月，当该网站开始投放广告以创收时，一些人抱怨，但该网站仍然很受欢迎。⁴⁷该网站早期得到了 Netscape 的支持，后者在自己的网站上添加了一个指向该网站的链接。当人们安装 Netscape 浏览器时，他们首先看到的是 Netscape 的主页，许多人从未费心去改变它。由于 Netscape 的管理层正在抨击微软，该公司直到为时已晚才意识到，它也发明了一个“门户”，只是他们自己不知道而已。Netscape 最终被 AOL 收购，后者是另一家擅长让普通人轻松进入复杂的网络空间的公司。

Besides the commercial sites mentioned above, another group of Web sites that emerged were search engines and “portals”: sites that help people navigate the Web and find information of interest to them. The most successful of these was Yahoo!, founded in 1994 as “Jerry’s Guide to the World Wide Web,” by Dave Filo and Jerry Yang while students at Stanford University. As the Web grew, a number of search engines appeared. Most used some sort of automated program (“bot,” short for “robot”) to “crawl” through Web sites and extract keywords. Yahoo! used automation, too, but its index was assembled by Yang and Filo— personally at first, later supplemented by other people. So from the onset Yahoo! was not just a search engine, but a place where a human touch took some of the fear out of visiting the Web naked. All of that was offered for free, and it remained so. When in August 1995 the site began running advertisements to generate revenue, a few complained but the site stayed popular.⁴⁷ The site got an early boost from Netscape, which put a link to it from its own site. When people installed the Netscape browser, Netscape’s home page was the first thing they saw, and many never bothered to change that. As Netscape’s management was railing against Microsoft, the company did not realize, until it was too late, that it had also invented a “portal” but did not know it. Netscape was eventually acquired by AOL, another company adept at easing ordinary people into the complexities of cyberspace.

精通计算机的互联网用户不需要门户网站。他们更喜欢强力搜索引擎，并且不怕使用布尔代数构建复杂的搜索来找到他们想要的内容。早期的领导者是 Altavista，由数字设备公司的硅谷实验室创立。Altavista 的成功不足以挽救其母公司，到 1998 年底，其搜索引擎被竞争对手谷歌超越，谷歌由谢尔盖·布林和拉里·佩奇（也在硅谷）创立。谷歌的成功在于它处理关键字检索的方式：它根据该网站被其他网站引用的次数对包含该词的网站进行排名——再次利用了网络空间的内置社区。

Computer-savvy Internet users did not need a portal. They preferred brute-force search engines and were not afraid to construct complex searches using Boolean algebra to find what they wanted. An early leader was Altavista, founded by the Silicon Valley lab of Digital Equipment Corporation. Altavista’s success was not enough to rescue its parent company, and by late 1998 its search engine was surpassed by a rival, Google, founded by Sergei Brin and Larry Page (also in Silicon Valley). Google’s succcess lay in the way it dealt with the retrieval of a keyword: it ranked sites containing that word according to how much that site was referenced by other sites—again taking advantage of the built-in community of cyberspace.

Google 的成功说明了网络的一个根本问题：在 20 世纪 90 年代末出现时，网络是有缺陷的。网络的创始人 Tim Berners-Lee 曾描述说，他的愿景只实现了一半——他最初想要一个既方便上网又方便输入信息的网络。48^用户应该能够构建自己的“门户”，完全满足他们的需求，雅虎也遵循了这一理念，推出了“我的雅虎！”功能，但这个功能远没有 Berners-Lee 设想的那样成为网络不可或缺的一部分。同样，那些我曾记录过的网络先驱——设想了“Xanadu”的 Ted Nelson 和创建了“在线系统”的 Doug Engelbart——如今都是网络最严厉的批评者。他们向越来越不愿意听的听众辩称，用户为超文本传输协议的扁平、单向文件结构付出了高昂的代价。⁴⁹几乎被遗忘的还有尤金·加菲尔德开发的科学信息检索系统，即科学引文索引。与谷歌一样，它通过记录有多少其他论文（如果有的话）在脚注中引用了它们来索引科学论文。引文索引在传统大型机上运行，并过渡到互联网时代，但它不再是帮助浏览知识空间的主要力量。WAIS 的创始人 Brewster Kahle 也认识到了网络的缺点，并于 2001 年推出了一个“互联网档案馆”，允许用户检索那些原本会消失在以太中的旧网站^。50

Google’s success illustrates something fundamental about the Web: as it emerged by the end of the 1990s, the Web is flawed. Tim Berners-Lee, its creator, described how only half of his vision has come true—he initially wanted a Web that was as easy to write to as it was to surf.⁴⁸ Users should be able to construct their own “portal” tailored precisely to their needs, a notion that Yahoo! has followed with its “My Yahoo!” feature but that is nowhere near as integral to the Web as Berners-Lee envisioned. Likewise, those I have chronicled as Web pioneers—Ted Nelson, who envisioned “Xanadu,” and Doug Engelbart, who created an “On Line System”—are today among the Web’s harshest critics. They argue, to an audience that increasingly does not want to listen, that users pay a high price for the flat, one-way file structure of the Hypertext Transfer Protocol.⁴⁹ Almost forgotten, too, was a system for the retrieval of scientific information developed by Eugene Garfield, called Science Citation Index. Like Google, it indexed scientific papers by noting how many other papers, if any, referenced them in their footnotes. Citation Index ran on traditional mainframes and made a transition to the Internet age, but it is no longer a major force in helping navigate thorough knowledge spaces. Brewster Kahle, the creator of WAIS, also recognized the Web’s shortcomings, and in 2001 he launched an “Internet Archive” that allowed users to retrieve old Web sites that have otherwise disappeared into the ether.⁵⁰

公地悲剧

Tragedy of the Commons

1968 年，在现代环保运动开始之际，生物学家加勒特·哈丁撰写了一篇名为《公地悲剧》的论文。在随后的几十年里，这篇论文被广泛引用，并被认为是对困扰工业社会的环境问题原因最有见地的观察之一^。51

In 1968, at the onset of the modern-day environmental movement, the biologist Garrett Hardin wrote a paper called “The Tragedy of the Commons.” It became widely cited and referred to in the decades that followed as one of the most insightful observations on the causes of the environmental problems afflicting industrial societies.⁵¹

在论文中，他将全球环境问题比作村庄的公共牧场，让“再多一只”动物放牧的成本很低，而该动物的主人获得的收益却很高。当总人口较少时，对公共牧场的损害微不足道。但随着放牧动物数量的增加，损害逐渐加剧，直到整个公共牧场被毁。在论文中，哈丁量化了个人利益与公共资源损害之间的经济关系，但他也认为，只有外部监管才能防止这种悲剧的发生。不受约束的市场力量和公共牧场使用者的利他行为都不足以解决问题。

In the paper he likened global environmental problems to a common grazing area of a village, where the cost of letting “just one more” animal graze is low, while the benefit to the owner of that animal is high. When the total population is low, the damage to the commons is insignificant. But as the number of grazing animals increases, the damage gradually becomes more severe, until the entire commons is destroyed. In the paper, Hardin quantifies the economic relationship of benefits to individuals versus damage to the common resource, but he also argues that nothing short of external regulation can prevent this tragedy from occurring. Neither unfettered market forces nor altruistic behavior by the commons’ users will suffice.

从严格的技术角度来看，互联网还远未达到人口过剩的程度。许多人，包括鲍勃·梅特卡夫，都曾预言互联网会在呈指数级增长时崩溃，但后来不得不食言。52^互联网经历了 1988 年蠕虫、病毒、千年虫危机、.com 崩盘和 2001 年 9 月 11 日恐怖袭击等挑战，几乎没有出现任何问题。它基于强大的设计。然而，就互联网所传递的信息内容和质量而言，它确实很糟糕。现已过时的 Gopher 界面提供的高质量文本文件的简单传递已经演变成充斥着弹出广告、垃圾邮件和色情内容的信息流。网上冲浪变得如此令人沮丧，以至于人们很难记得它曾经是多么令人兴奋。.com 的崩盘可能是一个信号，表明出现了问题，就像北大西洋渔业的崩溃一样。类似崩塌事件可能还会更多。如果约翰·佩里·巴洛 (John Perry Barlow) 等人有发言权，哈丁为自然环境提出的监管将很难在互联网上实施。美国在线时代华纳 (AOL Time Warner) 等几家大型媒体公司可能会进行整合和控制。这两种未来都不理想。第三种解决方案是可能的：保留互联网的基础设施，但按照纳尔逊、恩格尔巴特和伯纳斯-李等研究人员所倡导的思路重建其内容空间。53^这不是一项容易的任务，但互联网和万维网的建设本身也不容易。这是一个持续存在的问题，现在说如何解决或是否会解决还为时过早。

By a strict technical measure, the Internet has not come close to approaching this point of overpopulation. Many, including Bob Metcalfe, who predicted its collapse as it grew exponentially, have had to eat their words.⁵² The Internet passed through challenges like the 1988 worm, viruses, the Y2K crisis, the dot.com collapse, and the terrorists’ attacks of September 11, 2001, with hardly a hiccup. It is based on a robust design. As for the content and quality of information that the Internet conveys, however, it has indeed been tragic. The simple delivery of high-quality text files that the now-obsolete Gopher interface delivered has evolved into a stream of information polluted with pop-up ads, spam, and pornography. Web surfing has gotten so frustrating that one has a hard time remembering how exhilarating it once was. The collapse of the dot.coms may have been a signal that something was wrong, like the collapse of fisheries in the North Atlantic Ocean. More such collapses may be forthcoming. Regulation, which Hardin proposed for the natural environment, will not come easily to the Internet if people like John Perry Barlow have their say. Consolidation and control by a few giant media companies like AOL Time Warner may come to the fore. Neither future is appealing. A third solution is possible: leave the infrastructure of the Internet in place, but reconstruct its content space along the lines of what researchers like Nelson, Engelbart, and Berners-Lee have advocated.⁵³ That would not be an easy task, but neither was the construction of the Internet and the World Wide Web in the first place. This in an ongoing issue and it is too early to say how, or if, it will be resolved.

GNU/Linux

1988 年，已故的西摩·克雷 (Seymour Cray) 罕见地公开露面，描述了用砷化镓芯片制造新型超级计算机的计划。一位听众问他，使用这种不寻常的材料感觉如何，人们对其特性的了解远不如硅，据说克雷回答说：“好吧，如果你能发音，我想这就是使用它最困难的事情了。” ⁵⁴

In 1988 the late Seymour Cray made one of his rare public appearances, in which he described plans to build a new supercomputer with chips made from gallium arsenide. A member of the audience asked him what it was like working with this unusual material, whose properties were far less understood than those of silicon, to which Cray is said to have replied, “Well, if you can pronounce it, I guess that’s the hardest thing about working with it.”⁵⁴

这也适用于 Linux，自 1995 年以来，Linux 一直处于备受瞩目的运动的中心，旨在为微软的 Windows 提供替代操作系统，并打破 1990 年后威胁 UNIX 的方言之争。Linux 是 Linus Torvalds 的杰作，他于 1969 年出生于芬兰，他的名字发音为“Lee-noose”。这个名字在芬兰并不罕见，但他的父母以美国生物学家 Linus Pauling 的名字为他命名，而 Pauling 的名字（就像史努比的朋友一样）的 i 发音很长。至于操作系统本身，Linux 的追随者说它的发音是“Lih-nooks”，但许多人也将其发音为“Lie-nux”，或 i 发音短：“Lin-ux”。铁杆 Linux 粉丝会嘲笑任何发音“错误”的人，但由于 Linux 的吸引力之一是它是一个“开源”程序（稍后会详细介绍），因此他们目前只能排除新人。与这个难题相比，砷化镓的发音很容易^。55

That also applies to Linux, which since 1995 has been at the center of a highly visible movement to provide an alternative operating system to Microsoft’s Windows, and to break out of the Babel of dialects that was threatening UNIX after 1990. Linux is the creation of Linus Torvalds, born in 1969 in Finland, where his name is pronounced “Lee-noose.” The name is not rare in Finland, but his parents named him after the American biologist Linus Pauling, whose name (like Snoopy’s friend) is pronounced with a long i. As for the operating system itself, Linux adherents say it is pronounced “Lih-nooks,” but many also pronounce it “Lie-nux,” or with a short i: “Lin-ux.” Hard-core Linux fans scoff at anyone who pronounces it “wrong,” but since one of Linux’s appeals is that it is an “open source” program (more on that later), they can only exclude newcomers so far. Pronouncing gallium arsenide is easy compared to this conundrum.⁵⁵

暂且不谈这个问题，那么 Linux 是什么以及为什么人们对它如此感兴趣呢？稍微回顾一下 UNIX 的历史将会有所帮助。第 3 章讨论了操作系统的起源，最初由 IBM 大型机的首批客户开发，他们开发了“批处理”方法以使其昂贵的投资始终保持繁忙。第 7 章讨论了小型计算机的操作系统是如何根据不同的需求演变而来的。对于小型计算机和个人计算机，中央处理器和内存本身相对便宜，而且不需要一直让它们保持繁忙。对于这些机器，核心问题是信息从键盘进入处理器和主内存，从处理器到显示器和打印机，以及往返于外部（磁盘）存储器。UNIX 于 1969 年在贝尔实验室首次成型，是一套基本的文件管理工具，最初是为数字设备公司的 PDP-7 和后来的 PDP-11 小型计算机编写的。实际移动数据的命令是特定于机器的，但在 Ken Thompson 和 Dennis Ritchie 用编程语言“C”重写它之后，只要有人为这些机器编写了 C 编译器，UNIX 就可以移动（“移植”）到不同制造商的其他计算机上。而且由于 AT&T 的受监管垄断地位，UNIX 迅速传播到贝尔实验室以外的地区，尤其是大学，如前所述。

Leaving that issue unresolved for the moment, what is Linux and why all the interest in it? A little backtracking into the history of UNIX will be helpful here. chapter 3 discussed the origin of operating systems, initially by the first customers of IBM mainframes who developed “batch” methods to keep their costly investments busy all the time. chapter 7 discussed how operating systems for small computers evolved out of different needs. For minicomputers and personal computers, the central processor and memory itself were relatively cheap, and one did not need to always keep them busy. For these machines the central issue was the movement of information into the processor and main memory from keyboards, from the processor to monitors and printers, and to and from external (disk) storage. UNIX first took form at Bell Labs in 1969 as a set of basic file management tools, initially written for the Digital Equipment Corporation PDP-7 and later PDP-11 minicomputers. The commands that actually moved data were machine specific, but after Ken Thompson and Dennis Ritchie rewrote it in the programming language “C,” UNIX could be moved over (“ported”) to other computers from different manufacturers, as long as someone had written a C compiler for those machines. And because of AT&T’s position as a regulated monopoly, UNIX quickly spread to locations outside Bell Labs, especially to universities, as documented earlier.

从诞生之日起，UNIX 就具有了区别于其他所有编程系统的特征，这些特征对于理解 Linux 至关重要。其中一个特征是社交性的，而非技术性的：UNIX 在贝尔实验室以外的大学和其他非商业场所得到广泛使用。它的发展得益于分散各地的研究人员的合作努力，他们在现有版本的 UNIX 上进行改进 — — 因此成为早期的，也许是第一个“吃自己的狗粮”的例子。它的创造者 Ken Thompson 和 Dennis Ritchie 说：

From its origins, UNIX acquired characteristics that have set it apart from all other programming systems, and these characteristics are crucial to an understanding of Linux. One was social, not technical: the wide use of UNIX in universities and other noncommercial locations outside Bell Labs. It grew by a cooperative effort of researchers scattered widely, who worked with an existing version of UNIX as they built improvements—thus becoming an early, perhaps the first, instance of “eating your own dog food.” Ken Thompson and Dennis Ritchie, its creators, said that

如果系统设计者被迫使用该系统，他们很快就会意识到其功能和表面缺陷，并强烈希望在为时已晚之前纠正它们。由于所有源程序始终可用且易于在线修改，因此当其他人发明、发现或提出新想法时，我们愿意修改和重写系统及其软件^。56

if the designers of a system are forced to use that system, they quickly become aware of its functional and superficial deficiencies and are strongly motivated to correct them before it is too late. Because all source programs were always available and easily modified on-line, we were willing to revise and rewrite the system and its software when new ideas were invented, discovered, or suggested by others.⁵⁶

技术特征包括一个“内核”，它仅包含用于移动数据和控制处理器的最基本命令。UNIX 程序员能够通过组合简单的命令来构建复杂而精密的操作，主要是通过允许一个程序的输出数据作为另一个程序的输入数据（使用所谓的“管道”）。文件以最少的结构或格式存储在 UNIX 中：“系统不需要任何特殊的结构……文件的结构由使用它们的程序控制，而不是由系统控制。” ⁵⁷最后一个特征是系统对 C 编程语言的依赖和对 C 编译器的使用。使用 UNIX 的人都谈到了其初始设计的优雅。Ritchie 和 Thompson 谦虚地指出了他们必须使用的 PDP-7 的物理限制，这台计算机的性能大约相当于 Commodore VIC-20，Torvalds 的祖父小时候让他玩过其中的一台。他们显然对系统的设计进行了大量仔细的思考。⁵⁸

The technical characteristics include a “kernel” that consists of only the most essential commands to move data and control the processor. UNIX programmers are able to build up complex and sophisticated operations from combinations of simpler commands, primarily by allowing the output data of one program to serve as the input data for another (using so-called “pipes”). Files are stored in UNIX with a minimum of structure or formatting: “No particular structuring is expected by the system ... the structure of files is controlled by the programs that use them, not by the system.”⁵⁷ A final characteristic is the system’s reliance on the C programming language and the use of C compilers. Those who work with UNIX speak of the elegance of its initial design. Ritchie and Thompson modestly point to the physical constraints of the PDP-7 that they had to use, a computer whose power was about equal to a Commodore VIC-20, one of which Torvalds’s grandfather let him play with as a child. They clearly put a lot of careful thought into the system’s design.⁵⁸

UNIX 是在小型计算机上创建和培育的，并成为工作站的首选操作系统。在 1985 年之后销量更大的 IBM 兼容个人计算机上，情况有所不同。1980 年，在微软收购 IBM PC 的 DOS 之前，它开发了一个名为 XENIX 的 UNIX 版本。微软希望它能成为 16 位处理器的标准。59 ^XENIX销量不错，但行业朝着不同的方向发展。UNIX 的可移植性不那么重要，因为 IBM 架构将竞争对手赶出了个人电脑市场。而且 XENIX 比其他操作系统需要更多的内存。因此，个人电脑以西雅图计算机产品公司的 Tim Paterson 编写的简洁 DOS 为标准。DOS 的后续版本融合了 UNIX 的一些功能，例如文件排序的树状结构，甚至还有类似管道的东西，尽管不如 UNIX 提供的功能那么强大或优雅。其他功能，尤其是极为重要的多任务处理，要么从未出现在 DOS 中，要么不得不等到 Windows 中，而 Windows 将它们以一种不太优雅的方式移植了进来。DOS 存储文件的方式也不同，它对存储在文件中的数据施加了更多的结构。在其他方面，这两个系统有很多共同之处，尤其是用户在“命令行”中输入简短命令来移动、复制、删除或移植文件的方式。（2001 年秋天，微软正式淘汰了 DOS，推出了一个最终没有 DOS 代码作为核心的 Windows 版本。）

UNIX was created and nurtured on minicomputers and became the preferred operating system for workstations. On IBM-compatible personal computers, which after 1985 were being sold in greater numbers, the situation was different. In 1980, before Microsoft acquired a DOS for the IBM PC, it developed a version of UNIX called XENIX. Microsoft hoped it would become a standard for 16-bit processors.⁵⁹ XENIX sold well, but the industry evolved in a different direction. The portability of UNIX was less important, because the IBM architecture drove competitors from the personal computer market. And XENIX required more memory than other operating systems. Thus personal computers standardized around the spartan DOS, written by Tim Paterson of Seattle Computer Products. Later versions of DOS incorporated some of UNIX’s features, such as the treelike structure of ordering files, and even something like pipes although not as powerful or as elegant as what UNIX offered. Other features, especially the all-important multitasking, either never got into DOS or else had to wait for Windows, where they were grafted on in an inelegant fashion. The way DOS stored files was also different, imposing more structure on data stored in files. In other respects the two systems had much in common, especially in the way users typed in brief commands at a “command line” to move, copy, delete, or port files. (In the fall of 2001 Microsoft officially retired DOS when it introduced a version of Windows that, finally, did not have DOS code at its core.)

1991 年，时年 22 岁的 Torvalds 开始为他的 IBM 兼容个人计算机编写一个 UNIX 版本。60 他开始编写 UNIX 版本的^过程细节记录在他那段时期的回忆录中，这里对其进行了简要概述。前一年，他在赫尔辛基大学选修了一门 UNIX 课程。他说，老师和学生一样对这个学科不熟悉，但这不要紧，因为参加这门课程使他能够使用在分时 VAX 上运行的数字设备公司的 UNIX 版本。Torvalds 通过阅读阿姆斯特丹自由大学的 Andrew Tanenbaum 撰写的一本关于操作系统的书，对 UNIX 非常熟悉。Tanenbaum 还开发了一个教学版的 UNIX，他称之为“Minix”。⁶¹读过这本书，再加上童年时在 Sinclair QL 和 Commodore VIC-20 等个人电脑上进行过黑客攻击，托瓦兹确信他不仅想学习所有关于 UNIX 的知识，还想在家用电脑上使用 Minix。Sinclair 使用了先进的摩托罗拉处理器，但托瓦兹意识到 IBM-PC 标准已经非常成熟，他不得不改用基于英特尔的架构。就在那时，基于英特尔 80386 芯片的个人电脑开始出现在芬兰市场上——这种芯片使个人电脑能够提供与小型计算机和工作站相媲美的性能^。62

In 1991, Torvalds, then twenty-two years old, set out to write a version of UNIX for his IBM-compatible personal computer.⁶⁰ The details of how he came to do that are recorded in his memoir of the era, and are summarized briefly here. The previous year he had taken a course at the University of Helsinki on UNIX. The instructor, he said, was as new to the subject as the students, but that did not matter, because the enrollment in the course gave him access to a version of Digital Equipment Corporation’s UNIX running on a time-shared VAX. Torvalds was already familiar with UNIX from reading a book on operating systems by Andrew Tanenbaum, of Vrije University in Amsterdam. Tanenbaum had also developed a teaching version of UNIX, which he called “Minix.”⁶¹ Reading that book, combined with a childhood spent hacking on personal computers like the Sinclair QL and the Commodore VIC-20, convinced Torvalds that he wanted not only to learn all he could about UNIX, but also that he wanted Minix on his home computer. The Sinclair used an advanced Motorola processor, but Torvalds recognized that the IBM-PC standard was becoming so well established that he had to switch to the Intel-based architecture. Just at that time, PCs began to appear on the Finnish market based on the Intel 80386 chip—a chip that enabled personal computers to offer performance competitive with minicomputers and workstations.⁶²

托瓦兹 (分期付款) 购买了一台 386 电脑，大约一个月后，他购买了适用于该电脑的 Minix 版本。根据他的回忆录，1991 年冬天，他几乎刚安装并运行电脑，就发现 Minix 存在不足。具体来说，他需要一个终端仿真程序，这样他就可以使用电脑作为终端来访问大学的计算机及其软件资源并访问在线讨论组。63^他没有将其编写为 Minix 下的进程，而是编写了一个自己的仿真程序，用他的话来说，就是“裸机”编程。从那时起，最终出现了一个独立于 Minix 的 UNIX 版本，而且这个版本没有像他认为的 Minix 那样受到任何削弱或限制。 1991 年 7 月，他在新闻组上发布了一条简短的消息，向全世界暗示他正在考虑的不仅仅是一个扩展的终端仿真器，而且还要实现一个符合国际标准化组织下属委员会制定的标准的 386 UNIX 实现。64^一个人没有提供有关标准的信息，而是在赫尔辛基理工大学的一台计算机上提供了空间，托瓦兹可以在那里发布他的作品版本，其他人可以通过文件传输协议 (FTP) 下载。

Torvalds bought (on the installment plan) a 386-based PC, and a version of Minix for it about a month later. According to his memoir, almost as soon as he had it installed and running, in the winter of 1991, he found Minix wanting. Specifically, he needed a terminal-emulation program, which would allow him to use the PC as a terminal to access the university’s computer, with its software resources and access to online discussion groups.⁶³ Rather than write it as a process under Minix, he wrote an emulation program of his own, programming “to the bare metal” as he described it. From that beginning eventually came a version of UNIX that was separate from Minix, and that was not crippled or restricted in any way, as he felt Minix was. A brief note posted to a newsgroup in July 1991 gave a hint to the world that he was thinking not just of an expanded terminal emulator but of a 386 implementation of UNIX that would conform to a standard set out by a subcommittee of the International Standards Organization.⁶⁴ One person responded to his query not with information on the standard but with an offer of space on a computer at the Helsinki University of Technology, where Torvalds could post versions of his work and where others could download it via the file transfer protocol (FTP).

1991 年 8 月，他给讨论组的另一封简短说明进一步指出：“我正在为 386 (486) AT 克隆版开发一个 (免费) 操作系统 (只是业余爱好，不会像 gnu 那样庞大和专业)... 我希望听听大家对 minix 的喜好和厌恶的反馈，因为我的操作系统与它有点相似。” ⁶⁵请注意，他当时仍在考虑解决他在 Minix 中发现的缺陷。下个月，他在 FTP 站点上发布了自己的作品，在创建该站点的 Ari Lemke 的建议下，该程序被称为“Linux”。随着 Linux 逐渐独立，托瓦兹越来越不认为它是 Minix 的衍生品，而认为它是 UNIX 的新版本。这导致了他与 Tanenbaum 的决裂，Tanenbaum 在 1992 年初的新闻组帖子中表达了他对 Torvalds 做法的不满。Torvalds 强烈捍卫了他的工作，并伴随着当时在线讨论中常见的“激烈”言辞。66 ^（该讨论一直在“comp.os.minix”新闻组中进行；在与 Tanenbaum 激烈的争论之后，Linux 讨论转移到他们自己的新闻组中。）

Another brief note to the discussion group in August 1991 went further: “I’m doing a (free) operating system ( just a hobby, won’t be big and professional like gnu) for 386 (486) AT clones ... I’d like any feedback on things people like/dislike in minix, as my OS resembles it somewhat.”⁶⁵ Note that he was still thinking in terms of addressing the deficiencies he saw in Minix. The next month he posted his work on the FTP site, and at the suggestion of Ari Lemke, who made the site available, the program was called “Linux.” As Linux got legs of its own, Torvalds thought of it less and less as a derivative of Minix and more as a new version of UNIX. That led to a break with Tanenbaum, who registered his disapproval of Torvalds’s approach in a posting to the newsgroup in early 1992. Torvalds responded with a vigorous defense of his work, accompanied by “flaming” rhetoric that was typical of on-line discussions in those days.⁶⁶ (The discussions had been going on in the “comp.os.minix” newsgroup; after a heated exchange with Tanenbaum, Linux discussions moved to a newsgroup of their own.)

与此同时，托瓦兹在讨论列表上的人们的鼓励下继续致力于这个项目。就像 Netscape 的马克·安德森一样，托瓦兹利用互联网的能力，以低廉的价格在世界各地分发他的作品，并从热情的用户那里获得几乎即时的反馈。没有一个经典的营销理论能够预见到托瓦兹所建立的创作者和用户之间的密切关系。

Meanwhile Torvalds kept working on the program, with encouragement from people on the discussion list. Like Marc Andreesen at Netscape, Torvalds exploited the Internet’s ability to distribute his work cheaply around the world and to generate almost instant feedback from enthusiastic users. No classical theories of marketing could have foreseen the close relationship between creator and user that Torvalds developed.

尽管当时有各种各样的 UNIX 版本，但 Linux 满足了人们的需求。第 9 章中提到的 UNIX 的分裂状态对 Torvalds 有利。他的版本允许用户使用没有当时其他版本负担的操作系统。AT&T 曾希望在可以自由营销它之后从它的创造中获利，但是它的营销效果不佳。AT&T 坚持拥有“UNIX”名称，这确实惹恼了人们，导致出现了以“IX”结尾的其他名称、“UN*X”的拼写或其他愚蠢的说法。更糟糕的是，该公司起诉了销售包含 AT&T 代码（无论多小）的 UNIX 变体的供应商。由于美国政府通过 ARPA 的支持，伯克利 UNIX 发行版更容易获得，但就连 SUN 也为其后续几代工作站转向了封闭版本（Solaris）。 1993 年，AT&T 将其 UNIX 版本出售给 Novell，Novell 随后不久又将其转售。伯克利 UNIX 发行版已发展出多个版本，其中一些可免费使用（图 10.1）。

In spite of all the various flavors of UNIX then available, Linux was filling a need. The fragmentation of UNIX, mentioned in chapter 9, turned to Torvalds’s advantage. His version allowed users to get the operating system with none of the baggage associated with other versions then available. AT&T had hoped to profit from its creation once the company was free to market it, but its marketing was ineffective. AT&T did manage to annoy people with its insistence on owning the name “UNIX,” leading to other names ending in “IX,” the spelling “UN*X,” or other silliness. Worse than that, the company sued vendors who were selling variants of UNIX that had portions, however small, of AT&T’s code in them. The Berkeley distributions of UNIX were more accessible, owing to the U.S. government’s support through ARPA, but even SUN moved to a closed version (Solaris) for its later generations of workstations. In 1993 AT&T sold its version of UNIX to Novell, which resold it shortly thereafter. Berkeley distributions of UNIX have evolved into several versions, some of them freely available (figure 10.1).

在接下来的十年里，Linux 的规模和质量不断提高。围绕着它，出现了一批狂热者，对他们来说，Linux 不仅仅是一个操作系统。对他们来说，全球各地的志愿者成功开发 Linux 既是对弗雷德·布鲁克斯 (Fred Brooks) 在其经典著作《人月神话》 (1975) 中提出的软件工程原则的证明，也是对它的驳斥。回想一下，布鲁克斯写这本书是为了解释他作为项目经理在为 IBM 的 System/360 系列大型机开发操作系统时遇到的困难。他认为，在项目中期向软件开发团队增加人员只会让事情变得更糟，而不是更好，因为团队成员之间协调的困难压倒了新成员所能做出的任何贡献。对于越来越多的 Linux 爱好者来说，这似乎正是托瓦兹在不损害项目的情况下所做的：他征集了世界各地用户的帮助，并逐渐将 Linux 的部分内容移交给他信任的那些人。⁶⁷在他的书中，布鲁克斯还建议，大型软件项目应该由一位技术能力强、被赋予适当权力的领导者来组织。这当然适用于托瓦兹，尽管方式与布鲁克斯所建议的并不完全一样。托瓦兹没有任何正式的权力，但他能够以权威的言行来推动以他的名字命名的系统的演进。布鲁克斯认为，大型项目需要“制作人”和“技术总监”——在 Linux 开发中，艾伦·考克斯和托瓦兹分别担任这两个角色。68^无论如何，这个方法奏效了，而且随着托瓦兹从芬兰搬到硅谷，在一家名为 Transmeta 的公司工作，这个方法仍然有效。考克斯比托瓦兹大一岁，住在威尔士，但受雇于北卡罗来纳州的 Linux 公司 Red Hat。

For the next decade, Linux thus grew in size and quality. Around it grew a cadre of enthusiasts, for whom Linux was much more than just an operating system. To them, the successful development of Linux by dedicated volunteers working around the globe represented both a vindication and a refutation of the tenets of software engineering set out by Fred Brooks in his classic book The Mythical Man-Month (1975). Recall that Brooks wrote that book to explain the difficulties he had, as project manager, in developing an operating system for IBM’s System/ 360 line of mainframes. He argued that adding people to a software development team in the middle of a project makes things worse, not better, because the difficulties of coordination among its members overwhelm any contributions the new members can make. To the growing numbers of Linux enthusiasts, that seemed to be precisely what Torvalds was doing without harm to the project: he enlisted the help of users around the world, and gradually turned over portions of Linux to those among that group whom he trusted.⁶⁷ Elsewhere in his book Brooks suggests that one should organize a large software project around a leader with strong technical skills who is given proper authority. That certainly applies to Torvalds, even if not precisely in the way Brooks suggested. Torvalds has no authority in any formal sense, yet he is able to act and speak with authority on the evolution of the system that bears his name. Brooks argued that large projects need both a “producer” and “technical director”—roles that in Linux development are filled by Alan Cox and Torvalds, respectively.⁶⁸ In any event, it worked, and it continues to work as Torvalds moved from Finland to Silicon Valley, where he works at a company called Transmeta. Cox, a year older than Torvalds, lives in Wales but is employed by the Linux company Red Hat of North Carolina.

图 10.1

UNIX 车牌，复制品。原件归 DEC 员工、新罕布什尔州居民 Armando Stettner 所有，该州的州格言就印在车牌上。Stettner 是 DEC 的 UNIX 首席推广者，尽管他遭到了 Dave Cutler 的抵制，后者是专有 VMS 操作系统的架构师，后来又担任了 Microsoft 的 Windows NT 架构师。请注意必须注明贝尔实验室。史密森学会收藏，Eugene Miya 捐赠。

Figure 10.1

UNIX License Plate, replica. The original was owned by Armando Stettner, a DEC employee and resident of New Hampshire, whose state motto appears on its license plates. Stettner was the leading evangelist for UNIX at DEC, although he faced resistance from Dave Cutler, the architect of the proprietary VMS operating system and later of Microsoft’s Windows NT. Note the obligatory credit line to Bell Labs. Smithsonian Institution collections, gift of Eugene Miya.

Linux 的发展是对布鲁克斯定律的证明还是驳斥，对于 Linux 社区来说至关重要。弗雷德·布鲁克斯的作品受到尊重，但《人月神话》从根本上讲是一本关于失败项目的书。Linux 爱好者带着一个目的来阅读他的作品：证明他们不会像 IBM 那样失败。他们希望进一步表明，即使微软的方法与 IBM 不同，他们的工作将产生比微软更好的软件质量。布鲁克斯在 1995 年重新出版的书中增加了几章，详细讨论了支持和反对他关于团队如何开发大型软件项目的理论的证据。虽然他讨论了当时仍以“收缩包装”形式出售的商业 PC 软件的影响，但该版本是在 Linux 广为人知之前出版的。（布鲁克斯确实提到 UNIX 是一个“令人兴奋”的项目的例子，其品质源于有一个或几个专门的个人参与其创建。）⁶⁹

Whether the development of Linux is a vindication or refutation of Brooks’s law is of central importance to the Linux community. Fred Brooks’s writings are respected, but The Mythical Man-Month is fundamentally a book about a project that failed. Linux enthusiasts come to his writings with an agenda: to demonstrate that they will not fail as IBM did. They wish to show further that their work will produce better-quality software than what Microsoft can produce, even if Microsoft’s methods are different from IBM’s. Brooks added several chapters to the 1995 reissue of his book, discussing at length the evidence both for and against his theories of how teams develop large-scale software projects. Although he discussed the impact of commercial PC software, at that time still sold in “shrink-wrapped” packages, the edition was published before Linux became well known. (Brooks did mention UNIX as an example of an “exciting” project whose qualities stem from having one or a few dedicated individuals involved in its creation.)⁶⁹

Eric Raymond 是看到这种软件开发模式优点的程序员之一，他在一篇名为《大教堂与集市》的有影响力的文章中，结合 Brooks 等人关于大型软件项目的文章，对这种模式进行了分析。70 ^Raymond认为，通过让人们查看和修改源代码，可以发现并修复错误。这样做可以提高总体质量水平，速度比封闭系统快得多。正如 Raymond 所说（转述 Torvalds 的话），“只要有足够的眼光，所有错误都是浅显的。” ⁷¹ Raymond 的理念让人想起 20 世纪 70 年代的 Homebrew 计算机俱乐部的精神，当时共享是主流。它让人想起更早的 20 世纪 50 年代，当时 IBM 704 程序员联合起来成立了 SHARE，即使各个成员都为竞争的航空航天公司工作。

Eric Raymond was among the programmers who saw the merits of this model of software development, and, in an influential essay called “The Cathedral and the Bazaar,” he analyzed it with regard to writings by Brooks and others about large software projects.⁷⁰ Raymond argued that by letting people look at, and modify, the source code, bugs are found and fixed. As they do that, the general level of quality rises, much faster than it possibly could in a closed system. As Raymond said (paraphrasing Torvalds), “Given enough eyeballs, all bugs are shallow.”⁷¹ Raymond’s philosophy recalled the spirit of the Homebrew Computer Club of the 1970s, where sharing was the order of the day. It recalls an even earlier time, in the 1950s, when IBM 704 programmers banded together to form SHARE, even if the individual members worked for competing aerospace companies.

正如比尔盖茨在《致爱好者的公开信》（第 7 章）中反对 Homebrew 精神一样，微软也再次扮演了这一运动的敌人角色。在几次公开演讲中，微软高管抨击了开放源代码的理念，他们像盖茨在 1976 年所做的那样认为，“免费”软件不会适当地奖励有才华的程序员的辛勤工作。Linux 对消费者是免费的；谁来为它的开发买单呢？Linux 开发社区的大多数成员都有“日常工作”，他们的雇主允许他们开发 Linux 作为福利，希望他们能够继续开发商业产品。随着最近经济的衰退，其中一些雇主可能不那么慷慨了。其他免费软件由出版商 O'Reilly & Associates 支持，该公司通过销售有关 UNIX 的书籍赚钱。商业公司 VA Linux 也支持这项工作，但最近 VA Linux 宣布将将其业务模式转向专有产品（并将其名称更改为 VA Systems）。

Just as the Homebrew spirit was opposed by Bill Gates in his “Open Letter to Hobbyists” (chapter 7), so too did Microsoft again play the role of the enemy to this movement. In a few public speeches, Microsoft executives railed against the philosophy of making source code available, arguing as Gates did in 1976 that “free” software would not properly reward talented programmers for their hard work. Linux is free to the consumer; who pays for its development? Most members of the Linux development community have “day jobs,” where their employers allow them to work on Linux as a perk, in hopes of keeping their talents available for developing commercial products. With the recent downturn in the economy, some of those employers may be less generous. Other free software is supported by the publisher O’Reilly & Associates, which makes money selling books about UNIX. The commercial company VA Linux also supports this effort, but recently VA Linux announced it was shifting its business model to proprietary products (and changing its name to VA Systems).

微软高管还认为，没有什么可以阻止系统向十几个方向发展——只有托瓦兹及其核心合作者的辛勤工作和才华才能阻止这种情况的发生。如果没有托瓦兹和考克斯的奉献，Linux 能否成功仍是一个悬而未决的问题。微软某人于 1998 年撰写的一份内部备忘录被泄露给了埃里克·雷蒙德 (Eric Raymond)，后者立即将其发布在互联网上，后来被称为臭名昭著的“万圣节文件”。据称，在这份备忘录中，微软制定了计划，让 Windows 用户难以在他们的机器上安装或使用 Linux。72^至于托瓦兹对这一切的态度，他声称自己从未受到政治考虑的驱使。他也没有对微软感到任何敌意，只是认为微软的代码在技术上不如他的。无论如何，Linux 是靠着全球自愿爱好者团体的帮助而成长起来的。

Microsoft executives also argued that nothing would prevent the system from careening off in a dozen directions—a likely scenario that is only prevented by the hard work and talents of Torvalds and his inner circle of collaborators. Whether Linux can succeed without the dedication of Torvalds and Cox is an open question. An internal memo, written by someone at Microsoft in 1998, was leaked to Eric Raymond, who promptly posted it on the Internet, where it became known as the notorious “Halloween Document.” In it, Microsoft allegedly laid out plans to make it difficult for Windows users to install or use Linux on their machines.⁷² As for Torvalds’s attitude toward all this, he claimed that he was never motivated by political considerations. Nor did he feel any animosity toward Microsoft, other than believing its code was technically not as good as his. In any event, Linux grew on the shoulders of a worldwide, voluntary group of enthusiasts.

GNU

任何有关 UNIX 的讨论都必须包括对 C 编程语言的讨论，该语言是与 UNIX 同步开发的，UNIX 是用 C 语言重写的。因此，1991 年秋天，托瓦兹做的第一件事就是在他的计算机上使用 C 编译器。这不是一项简单的任务，但一旦完成，他就可以访问 C 程序库，这使他能够引导自己进入更发达的 UNIX 系统。他选择的编译器是 GCC，即“GNU C 编译器”，由马萨诸塞州剑桥的 Richard Stallman 编写^。73

Any discussion of UNIX must include a discussion of the C programming language, which was developed in tandem with it, and in which UNIX was rewritten. Thus one of the first things Torvalds did, in the fall of 1991, was to adapt a C compiler for use on his machine. It was not an easy task, but once it was done he would gain access to a library of C programs, which allowed him to bootstrap his way to a more developed UNIX system. The compiler he chose was GCC, the “GNU C Compiler,” which had been written by Richard Stallman of Cambridge, Massachusetts.⁷³

托瓦兹与斯托曼的关系就像他与塔南鲍姆的关系一样复杂。托瓦兹依赖老程序员的工作成果，但他故意朝着与他们两人都不喜欢的方向出发。然而，就 C 编译器而言，斯托曼的编译器即使不是最好的，也是最好的编译器之一。它的广泛采用为 C 编程语言建立了事实上的标准 — 也就是说，标准 C 被定义为 GNU C 编译器可以理解的语言。而且它是免费的。非常免费。

Torvalds’s relationship with Stallman, like his relationship with Tanenbaum, is complex. Torvalds relied on work done by the older programmers, but he deliberately set off in a direction different from what either of them would have preferred. Nevertheless, in the case of a C compiler, Stallman’s was one of the best available, if not the best. Its widespread adoption established a de facto standard for the C programming language—that is, a standard C was defined as a language that the GNU C Compiler understood. And it was free. Very free.

斯托曼致力于开发和传播以精心定义的方式“自由”的软件，这一点已广为人知。74^他的冒险经历始于麻省理工学院人工智能实验室，这里是 PDP-10 和“不兼容分时系统”的所在地（第 7 章）。20 世纪 80 年代初，他认为，随着一家公司成立，试图将实验室中进行的部分人工智能研究商业化，实验室的道德规范（程序员自由分享其工作成果）正在被逐渐瓦解。斯托曼决心开发类似的软件，但不同之处在于，软件将免费赠送。将人工智能研究商业化的商业模式本来就是有缺陷的（尤其是在 1985 年左右 ARPA 撤回资金支持之后），但斯托曼的决心没有动摇。他决定创建一个 UNIX 版本（一个他知之甚少的操作系统），然后免费赠送。⁷⁵他本来希望复制 PDP-10 上使用的系统 ITS，但 UNIX 的优势在于它不止在一家制造商的一台或两台机器上使用。他将自己的 UNIX 副本称为“GNU”（发音为双音节词）：这是“Gnu's Not UNIX”的递归缩写。

Stallman’s campaign to develop and spread software that is “free” in a carefully defined way is by now well known.⁷⁴ His odyssey began at the MIT Artificial Intelligence Laboratory, the home of a PDP-10 and the “Incompatible Timesharing System” (chapter 7). In the early 1980s he felt that the ethic of the lab, in which programmers freely shared their work, was being drained away by the founding of a company that sought to commercialize some of the AI research being done there. Stallman resolved to produce similar software, with the distinction that it would be freely given away. The business model of commercializing AI research was flawed anyway (especially after ARPA withdrew financial support around 1985), but Stallman’s resolution remained. He decided to create a version of UNIX, an operating system he knew little about, and give that away.⁷⁵ He would have preferred copying ITS, the system used on the PDP-10, but UNIX had the advantage of being used on more than just one or two machines from a single manufacturer. He called his UNIX copy “GNU” (pronounced as a two-syllable word): a recursive acronym for “Gnu’s Not UNIX.”

1983 年 9 月，他在 Usenet 上发帖称：“从今年感恩节开始，我将编写一个完整的 Unix 兼容软件系统，称为 GNU（Gnu's Not Unix 的缩写），免费提供给所有可以使用的人。我们非常需要大家贡献时间、金钱、程序和设备。” 在同一篇帖子的后面，他阐述了他对自由软件的理念：“我认为黄金法则要求，如果我喜欢一个程序，我必须与喜欢它的人分享。我不能心安理得地签署保密协议或软件许可协议。为了能够继续使用计算机而不违背我的原则，我决定整合足够多的自由软件，这样我就可以不用任何非自由软件了。” ⁷⁶他对“自由”的定义会发生变化，但始终坚持自由软件永远不会与带有限制的软件纠缠在一起的理念。

In a Usenet post in September 1983 he wrote: “Starting this Thanksgiving I am going to write a complete Unix-compatible software system called GNU (for Gnu’s Not Unix), and give it away free to everyone who can use it. Contributions of time, money, programs and equipment are greatly needed.” Later in the same posting he stated his philosophy of free software: “I consider that the golden rule requires that if I like a program I must share it with other people who like it. I cannot in good conscience sign a nondisclosure agreement or a software license agreement. So that I can continue to use computers without violating my principles, I have decided to put together a sufficient body of free software so that I will be able to get along without any software that is not free.”⁷⁶ His definition of “free” would change but would always retain the notion that free software would never become entangled with software that had restrictions attached to it.

Stallman 发现，UNIX 系统的某些组件已经可以免费使用，其中包括 X-Windows 系统和 Donald Knuth 的 TeX 排版程序。除此之外，他还添加了自己在 MIT 时编写的 Emacs 文本编辑系统，并将其重写为 UNIX。当然还有 C 编译器。77^为了避免与 AT&T 产生法律纠纷，他拒绝查看 AT&T UNIX 的任何源代码，就像那些为 IBM PC 重新创建 BIOS 的人在“洁净室”工作一样。因此，GNU C 编译器 (gcc) 被提供作为“pcc”的替代品，后者是由 AT&T 的 Steve Johnson 编写的，并有偿提供。Stallman 的“Bison”解析器生成器（编译器的一个组件）被提供作为“yacc”（“又一个编译器”）的免费替代品，后者也是由 Steve Johnson 编写的。请注意 yak 一词的双关语；Stallman 遵循了黑客的古老传统，只要有可能就会使用双关语。⁷⁸他自己完成了大部分工作，在麻省理工学院慷慨借给他的办公室里。有一段时间，这也是他的卧室。（与 Linux 一样，GNU 也受益于那些有权使用计算资源的人非正式但真实的慷慨。）斯托曼的手因不断重复的击键而疼痛。在几位同事的帮助下，他开始构建一个软件系统，其软件系统可与商业或学术界大型团队创建的软件系统相媲美，甚至更好。

Stallman found that some components of a UNIX system were already available free—the X-Windows system and Donald Knuth’s TeX typesetting program, among others. To those he added the Emacs text editing system he had already written while at MIT, which he rewrote for UNIX. And of course the C compiler.⁷⁷ To avoid legal problems with AT&T, he refused to look at any source code for AT&T UNIX, just as those who recreated the BIOS for the IBM PC worked in a “clean room.” Thus the GNU C Compiler (gcc) was offered as a replacement for “pcc,” a compiler written by AT&T’s Steve Johnson and offered for a price. Stallman’s “Bison” parser generator (a component of a compiler) was offered as a free replacement for “yacc” (“yet another compiler compiler”), also written by Steve Johnson. Note the pun on the word yak; Stallman follows an old hacker tradition of punning whenever possible.⁷⁸ He did much of this himself, in an office generously loaned to him by MIT. For a while that was also his sleeping quarters. (As with Linux, GNU benefited from an informal but nevertheless real generosity from those who had access to computing resources.) Stallman’s hands suffered from the stress of constant, repetitive keystroking. With the help of a few colleagues, he began building up a body of software comparable to, or even better than, what it took large teams of people to create in the commercial or academic world.

尽管编程很重要，但斯托曼为保证其作品的权利而制定的法律协议也同样重要。在律师的帮助下，他制定了“GNU 通用公共许可证”（GPL），不仅将他的作品放入公共领域，还要求使用和修改它的人也将他们的修改放入公共领域。用他的话来说，“这是一个法律文书，要求那些传递程序的人拥有使用、修改和重新分发代码的权利；代码和自由在法律上是不可分割的。” ⁷⁹它不会阻止某些人像 Red Hat 和 VA Linux 等公司那样出售代码以牟利。但是，它确实阻止他们拥有他们所销售产品的代码部分。⁸⁰

As important as that programming was, just as important was the legal agreement Stallman crafted to guarantee the rights to his work. With the help of an attorney he developed a “GNU General Public License” (GPL) that not only put his work into the public domain, it also required that those who used and modified it put their modifications in the public domain as well. In his words, “It is a legal instrument that requires those who pass on a program to include the rights to use, modify, and redistribute the code; the code and the freedoms become legally inseparable.”⁷⁹ It does not prevent someone from selling the code for a profit, as companies like Red Hat and VA Linux do. It does, however, prevent them from owning the code portion of what they sell.⁸⁰

最后一条规定非常激进，它定义了 Linux（Torvalds 在 GPL 下发布）和其他所谓开源软件的特征。使用“免费”软件并不是什么新鲜事：微软从 BASIC 编程语言开始，该语言由达特茅斯开发，但由盖茨、艾伦和大卫多夫修改。BASIC 的创建者希望他们的程序能够广泛传播，并公开描述了其规范。微软为 Altair 开发的 BASIC 版本与达特茅斯的 BASIC 截然不同，这是 Kemeney 和 Kurtz 不赞成的，但对现有语言进行此类修改却很常见。其他类型的个人计算机软件，包括 dBase II 和一些文字处理器，都是对大学或政府支持开发的软件的修改，因此被认为是“免费的”。但一旦修改，它们的创建者就可以出售它们以牟利，而且确实这样做了。Linux、GNU 工具或 GPL 下的其他软件不能这样做。此外，GPL 还规定，如果在产品中使用此类软件，则整个产品都必须受 GPL 约束，即使产品的其他部分之前是专有的。因此，GPL 将专有软件“转化”为自由软件，这与业界的情况正好相反。正是这一规定对微软等公司构成了威胁，并导致微软的高管谴责这一运动违背了自由企业的“美国方式” ^。81

It is this last provision that is so radical, and it defines the character of Linux (which Torvalds released under the GPL) and other so-called open source software. Using “free” software was nothing new: Microsoft began with the BASIC programming language, developed at Dartmouth but modified by Gates, Allen, and Davidoff. The creators of BASIC wanted their program to become widespread and publicly described its specifications with that in mind. Microsoft’s version of BASIC for the Altair was very different from Dartmouth’s BASIC, something that Kemeney and Kurtz did not approve of, but such modifications of existing languages were common. Other types of personal computer software, including dBase II and some word processors, were modifications of software developed at universities or with government support and therefore considered “free.” But once modified, their creators could and did sell them for a profit. One cannot do that with Linux, with GNU tools, or with other software under the GPL. What is more, the GPL requires that if one uses such software in a product, the entire product must be covered by the GPL, even if other parts of it had previously been proprietary. The GPL thus “converts” proprietary software into free software, the opposite of what had been happening in the industry. It is this provision that is so threatening to companies like Microsoft, and that led senior executives at Microsoft to denounce the movement as being contrary to the “American Way” of free enterprise.⁸¹

许多关于 Linux 历史的流行记述都强调了 Torvalds 和 Stallman 之间的分歧，他们年龄只相差 16 岁，但在很多方面却代表了不同的时代。这些记述忽略了一个事实，即 Torvalds 完成了 Stallman 着手做的事情，而且 Linux 受到 GPL 的保护。Stallman 在开发 UNIX 内核时遇到了困难，而 UNIX 内核是 Torvalds 编写的第一款产品。Stallman 来自 DEC 大型机（如 PDP-10）和小型机（如 PDP-11）的环境；他可能没有意识到英特尔 80386 和后续芯片是如何占据主导地位的。或者可能有其他原因，与如何编写好内核的基本理念有关。无论如何，Stallman 提醒人们，全套 UNIX 工具应该称为“GNU/Linux”，而不仅仅是“Linux”。在这一点上他是对的。一些 GPL 哲学的拥护者创造了“开源”一词，主要是为了与他个人保持距离，但属于公共领域但不受 GPL 或其同等条款保护的软件不属于这一社会现象。

Many popular accounts of the history of Linux emphasize the rift between Torvalds and Stallman, who are only sixteen years apart in age but represent different generations in many ways. Such accounts neglect the fact that Torvalds completed what Stallman set out to do, and Linux is protected by the GPL. Stallman ran into difficulties in developing a UNIX kernel, which was the first thing Torvalds wrote. Stallman came from an environment of DEC mainframes like the PDP-10 and minicomputers like the PDP-11; he may not have recognized how the Intel 80386 and successor chips were taking over. Or there may have been other reasons, related to the underlying philosophy of how to write a good kernel. In any event, Stallman reminds people that the full set of UNIX tools should be called “GNU/Linux,” not just “Linux.” On that he is correct. Some advocates of the GPL philosophy coined the term open source, mainly to distance themselves from him personally, but software that is in the public domain but not covered by the GPL or its equivalent is not a part of this social phenomenon.

如果不是万维网和 Netscape 浏览器诞生后互联网活动的激增，这种活动可能就不会被注意到。突然间，人们需要更大的计算机来运行网站和路由互联网流量。UNIX 完全符合这一需求。除了 Linux 之外，Berkeley UNIX 的免费版本（其中所有 AT&T 代码都经过“清理”）也成为了首选的 UNIX 系统。每个版本都为 Web 管理员提供了修改和扩展系统的能力，而无需承担法律麻烦或费用。截至撰写本文时，最流行的 Web 服务器软件是 Brian Behlendorf 协调的 Apache（“补丁服务器”）；最流行的邮件路由程序是 Eric Allman 编写的“Sendmail”；网页最常用的脚本语言是 Larry Wall 编写的 Perl。所有这些都基于 UNIX，并且都是免费的。像 Google 这样的 Web 巨头依赖 Linux，而如上所述，Microsoft 的 Hotmail 服务运行在免费的 Berkeley 发行版上。

This activity might have gone unnoticed had it not been for the explosion of Internet activity after the creation of the World Wide Web and the Netscape browser. Suddenly there was a demand for larger computers to run Web sites and route Internet traffic. UNIX fit that need perfectly. Besides Linux, a free version of Berkeley UNIX, from which all the AT&T code was “cleansed,” became the UNIX systems of choice. Each offered Web administrators the ability to modify and extend their systems without legal troubles or charges. As of this writing, the most popular Web server software is Apache (“a patchy server”), coordinated by Brian Behlendorf; the most popular mail-routing program is “Sendmail,” written by Eric Allman; and the most-used scripting language for Web pages is Perl, written by Larry Wall. All are based around UNIX, and all are free. Web giants like Google rely on Linux, and as mentioned, Microsoft’s Hotmail service runs on the free Berkeley Distribution.

IBM

苹果公司 1984 年的超级碗广告宣称，未来与奥威尔的设想完全不同，但苹果公司在某一方面错了。奥威尔预言，今天的敌人将成为明天的盟友，而随着这种情况的发生，所有关于以前结盟或战争的记忆都将被抹去。在奥威尔的小说中，这就是温斯顿·史密斯的工作：将所有以前结盟的记录放入“记忆洞”。2001 年，超级碗广告中隐含的敌人 IBM 成为了盟友；微软成为了敌人。行业媒体尽职尽责地完成了温斯顿·史密斯的工作，抹去了每个人对二十年前这两个实体相对角色的记忆。

Apple’s 1984 Super Bowl commercial announced that the future would be nothing like Orwell’s vision, but in one respect Apple was wrong. Orwell predicted a world where today’s enemy becomes tomorrow’s ally, and as that happened, all memory of the previous alliance or warfare was erased. In Orwell’s novel, that was Winston Smith’s job: to put all records of previous alliances down the “memory hole.” In 2001 IBM, the implied enemy in the Super Bowl commercial, became an ally; Microsoft became the enemy. The trade press dutifully does Winston Smith’s job of erasing everyone’s memory of the relative roles of those two entities twenty years ago.

IBM 已将 Linux 视为封闭式专有操作系统的替代品，在“大型计算机”时代，IBM 曾对其严加保护。在 IBM 成为计算机行业的霸主期间，IBM 一直以制定自己的标准（例如 EBCDIC 而不是 ASCII）而闻名，并试图将这些标准强加给计算机界的其他成员。但 IBM 在 20 世纪 90 年代初遭受了巨大损失，如果不做出改变，它可能会与 BUNCH 的其他成员一起消失或变得无足轻重。该公司拥抱 Linux 的时间太短，很难形成一个连贯的故事，但它似乎始于 IBM 位于德国斯图加特郊外博布林根实验室的一个“臭鼬工厂”项目，1999 年底，一群年轻的程序员成功地将 Linux 移植到 IBM 390 大型机上。但这并不是 IBM 第一次接触开源软件：在 1996 年亚特兰大夏季奥运会上，IBM 是提供计算机服务（包括网站展示）的主要承包商。它没有“自食其力”，而是选择了 Apache，而不是使用 IBM 子公司 Lotus 提供的服务器。82 当时^IBM陷入了财务困境，它知道全世界都会关注它如何出色地应对奥运会，因此它对 Apache 的信心要大于对 Lotus 产品（Domino）的信心。

IBM has embraced Linux as an alternative to the closed, proprietary operating systems it so jealously guarded in its days of “big iron.” Throughout its reign as the dominant computer company, IBM was known as a company that set its own standards—think of EBCDIC instead of ASCII—and tried to impose them on the rest of the computing community. But IBM suffered huge losses in the early 1990s and might have vanished or shriveled into insignificance, along with the rest of the BUNCH, had it not changed. The company’s embrace of Linux is too recent for a coherent story to emerge, but it appears that it began as a “skunk works” project at IBM’s lab in Bo¨blingen, Germany, outside Stuttgart, where in late 1999 a team of young programmers succeeded in porting Linux to an IBM 390 mainframe. That was not IBM’s first exposure to open source software, though: for the 1996 Summer Olympics in Atlanta, IBM was the prime contractor in providing computer services, including a Web presence. Instead of “eating your own dog food” and using a server offered by the IBM subsidiary Lotus, it chose Apache instead.⁸² Financially troubled, IBM knew that the world would be watching how well it handled the Olympics, and it had more faith in Apache than in the Lotus product (Domino).

IBM 对“臭鼬工厂”开发非主流产品的想法也并不陌生。IBM 一直支持此类活动，至少可以追溯到 Thomas Watson, Jr. 著名的“野鸭”备忘录（第 9 章）。令人吃惊的是，这项工作如何通过了坚持传统 IBM 方式的各层管理人员和程序员的审核，直到获得董事长 Lou Gerstner 的关注和批准。据一份报告称，1998 年 12 月，纽约时报记者 John Markoff 报道了IBM计划将其一名员工开发的邮件程序作为开放源代码发布，此后Linux 工作陷入了备忘录、会议和委员会的漩涡中^。83据称，Gerstner 读了这篇报道，并要求 IBM 从那天起制定连贯的开放源代码政策。无论如何，IBM 对 Linux 做出了重大承诺，并宣布 2002 年将投入 20% 的研发预算用于在其产品线上运行 Linux。⁸⁴在行业媒体的一系列广告中，IBM 宣布：“事实显而易见：Linux 已经到来，Linux 已做好准备。为商业做好准备。为电子商务做好准备。为企业做好准备。”其中一条广告展示了一只模糊的黑白企鹅图像，企鹅是 Linux 的吉祥物，它正在大型机安装的塔楼间穿行，广告词为“事实：Linux 在企业中”。

Nor was the notion of a skunk works developing something out of the mainstream all that foreign to IBM either. It had always supported such activities, going back at least to the famous “wild duck” memo from Thomas Watson, Jr. (chapter 9). What was startling is how this effort made its way through the layers of managers and programmers committed to the classic IBM way, until it gained the attention and approval of the chair, Lou Gerstner. According to one account, the Linux effort was stuck in a vortex of memos, meetings, and committees when John Markoff, a reporter for the New York Times, reported in December 1998 on IBM’s plan to release a mail program developed by one of its staff as open source.⁸³ Gerstner allegedly read the story and demanded that IBM develop a coherent policy on open source from that day onward. In any event, IBM made a substantial commitment to Linux and announced that for the year 2002 it would devote 20 percent of its R&D budget toward getting Linux to run on its product line.⁸⁴ In a series of ads in the trade press, IBM announced: “The facts are clear: Linux is here and Linux is ready. Ready for business. Ready for e-business. Ready for enterprise.” One ad showed a fuzzy black-and-white image of a penguin, the Linux mascot, walking through the towers of a mainframe installation, with the caption “Fact: Linux in the enterprise.”

也许 IBM 之所以选择 Linux，是因为它在与微软和 Windows 打交道方面经验丰富，而它自己的个人计算机操作系统 OS/2 却鲜有问津。从多数观点来看，微软在这笔交易中占了上风，而 IBM 大型程序员团队历经数月研发的 OS/2 则被遗忘了。或者原因可能仅仅是 Linux 提供了最好的功能集。这个故事还在继续，目前还不清楚 Linux 是否会占上风，但自从这些广告出现以来，IBM 更进一步，推出了一系列只运行 Linux 的计算机。85 ^IBM在新闻稿中声称，基于每秒数百万条指令，Linux 系统占其 2001 年底交付的计算能力的 11%。而这一切都发生在 Torvalds 开始为他的 PC 编写终端仿真器十年之后。

Perhaps the reason for this embrace of Linux was IBM’s experience with Microsoft and Windows, versus its own personal computer operating system, OS/2. By most accounts, Microsoft got the best of that deal, and OS/2, the product of months of work by a large team of IBM programmers, was forgotten. Or perhaps the reason was simply that Linux offered the best set of capabilities. This story is very much ongoing, and it is not at all clear that Linux will prevail, but since those ads appeared IBM has gone even further and introduced a line of computers that run only Linux.⁸⁵ In the press release IBM claimed that Linux systems account for 11 percent of the computing capacity that it shipped in late 2001, based on millions of instructions per second. And all of this happening only ten years after Torvalds began writing a terminal emulator for his PC.

结论

Conclusion

即使人们学会了如何发音，西摩·克雷也从未成功制造出可以与硅竞争的砷化镓电路。无论如何发音，Linux 都必须以某种方式与微软打交道。Java 的经验表明，仅仅成为微软的替代品本身并不足以获胜。Linux 的宣传者中有一群强大而直言不讳的人，他们吹捧基于 Linux 的程序，这些程序提供与 Windows（“KDE”和“GNOME”）、文字处理器（“AbiWord”）和其他产品类似的图形界面。86^与Windows 不同，为了遵循 UNIX 理念，生成图形用户界面的代码与基本 Linux 代码分开。Linux 仍然可以通过键入命令行来访问，就像 DOS 一样。随着微软放弃 DOS，Linux 爱好者坚定地喜欢输入神秘的命令，其中许多命令类似于旧的 DOS 命令。苹果似乎对此持两种态度。当苹果于 1984 年推出 Macintosh 时，它取消了命令行，但随着最新版本的 Mac 操作系统（“X”，基于 UNIX），精明的用户可以绕过苹果公司闻名遐迩的图形界面。

Seymour Cray never did succeed in building gallium arsenide circuits that could compete with silicon, even if people did learn how to pronounce the term. Linux, however it is pronounced, is going to have to deal with Microsoft one way or another. The experience with Java shows that simply being an alternative to Microsoft is not sufficient in itself to prevail. Among Linux evangelists are a strong and vocal group who tout Linux-based programs that offer a graphical interface like Windows (“KDE” and “GNOME”), word processors (“AbiWord”), and other products.⁸⁶ In keeping with the UNIX philosophy, and in contrast to Windows, the code that generates the graphical user interfaces is kept separate from the base Linux code. Linux is still accessed by typing a command line, like DOS. As Microsoft moved away from DOS, Linux enthusiasts steadfastly prefer typing cryptic commands, many of which resemble the DOS commands of old. Apple seems to be of two minds on this. When it introduced the Macintosh in 1984 it got rid of the command line, but with the latest version of the Mac operating system (“X,” based on UNIX), a savvy user can bypass the graphical interface that Apple made so famous.

“Slashdot.org”网站每天都会发布与微软的斗争信息，其语气表明，口水战毕竟还没有结束。但并非所有信息都支持这种方法：Slashdot 的创始人之一、网名“CmdrTaco”的罗布·马尔达 (Rob Malda) 最近采取了不那么激进的立场。曾在 Red Hat 工作的拉斯·米切尔 (Russ Mitchell) 则持更怀疑的态度。他认为，与微软正面交锋是浪费时间；微软已经赢得了这场战斗。他希望看到 Linux 像 IBM 一样在服务器市场建立更强大的地位。如果执行得当，微软可能无法威胁到该市场。87^就微软而言，它不会让这种情况不战而降。

The Web site “Slashdot.org” posts daily messages on the battle against Microsoft, the tone of which suggests that flame wars are not extinct after all. But not all messages are in favor of this approach: Rob Malda, one of the founders of Slashdot who goes by the screen name “CmdrTaco,” recently took a less aggressive stance. And Russ Mitchell, who worked for Red Hat, is even more skeptical. He argues that going against Microsoft head to head is a waste of time; Microsoft has won this battle. He hopes to see Linux establish a stronger position in the server market, as IBM has done. If well executed, Microsoft might be unable to threaten that market.⁸⁷ For its part, Microsoft is not going to let this happen without a fight.

Linux 的拥护者们或许应该从马克·安德森的经历中吸取教训，当时他吹捧 Netscape Navigator 是 Windows 的竞争对手。在一次采访中，他将 Windows 描述为“一组部分调试的设备驱动程序”。88比尔·盖茨和史蒂夫^·鲍尔默并不认为这很有趣。如今，Netscape 被埋在美国在线的一个角落里。89^安德森和吉姆·克拉克都没有直言不讳地解释为什么 Netscape 最终在浏览器大战中输给了微软，但这种傲慢的言论并没有起到什么帮助作用。应该有人提醒安德森一句民间智慧：“不要拉超人的斗篷。”除非你是 IBM。无论如何，最初作为微软反垄断审判的注脚，林纳斯·托瓦兹声称“这只是一种爱好，不会成为大事或专业”，最终却变得相当有趣。我们拭目以待。

Linux evangelists might learn from the experience of Marc Andreesen, when he was touting Netscape Navigator as a competitor for Windows. In an interview he described Windows as “a partially-debugged set of device drivers.”⁸⁸ Bill Gates and Steve Ballmer did not think that was funny. Today, Netscape is buried in a corner of America Online.⁸⁹ Neither Andreesen nor Jim Clark has been forthright about why Netscape ultimately lost the browser war to Microsoft, but the hubris of statements like that one did not help. Someone should have reminded Andreesen of the folk wisdom, “You don’t tug on Superman’s cape.” Unless you are IBM. In any event, what started out as a footnote to the Microsoft antitrust trial, something that Linus Torvalds claimed was “just a hobby, won’t be big and professional,” is turning out to be quite interesting, after all. We will see.

结论：世界图景的数字化

Conclusion: The Digitization of the World Picture

1945 年至 2002 年间，计算机一次又一次地自我改造，每次都重新定义其本质。“计算机”最初是一种快速科学计算器；埃克特和莫奇利将它改造成通用数据处理机器 UNIVAC。肯·奥尔森将它改造成与用户协同工作的实时信息处理器。埃德·罗伯茨将它改造成任何人都可以拥有和使用的设备。史蒂夫·乔布斯和史蒂夫·沃兹尼亚克将它改造成既有用又有趣的设备。加里·基尔多尔和威廉·盖茨将它改造成一个标准化平台，可以运行零售店出售的大量商业软件。鲍勃·梅特卡夫、蒂姆·伯纳斯-李等人将它改造成一个通往全球网络的窗口。

Between 1945 and 2002 the computer transformed itself over and over again, each time redefining its essence. “The computer” started out as a fast scientific calculator; Eckert and Mauchly transformed it into UNIVAC, a machine for general data processing. Ken Olsen made it into a real-time information processor that worked symbiotically with its users. Ed Roberts transformed it into a device that anyone could own and use. Steve Jobs and Steve Wozniak turned it into an appliance that was both useful and fun. Gary Kildall and William Gates transformed it into a standardized platform that could run a cornucopia of commercial software sold in retail stores. Bob Metcalfe, Tim Berners-Lee, and others turned it into a window to a global network.

每次转型都伴随着断言，即进一步转型不太可能发生，但每次总有人成功突破。最近的转型，即向万维网转型，也曾有言论称计算机行业停滞不前，套用一位软件销售员的话来说，“再也没有唾手可得的果实了”。他错了，那些预测万维网是计算的最终归宿的人无疑也是错的。

Each transformation was accompanied by assertions that further transformations were unlikely, yet each time someone managed to break through. The latest transformation, to the World Wide Web, was also preceded by statements that the computer industry was stagnating, that there was, to paraphrase a software salesman, “no more low-hanging fruit.” He was wrong, and those who predict that the World Wide Web is the ultimate resting place for computing will no doubt be wrong as well.

到 20 世纪 90 年代中期，个人计算机已成为一种商品，商业软件因此成为向用户传达创新的中心。软件分层的过程始于为 UNIVAC 开发的第一个“自动编码”方案，并一直持续至今。这是扩大市场以包括没有编程倾向或才能的用户的唯一方法。同样，随着软件的每一次进步，人们听说“编程的终结”已经到来，现在“任何人”都可以让计算机做他或她想做的事。随着新市场的开放，编程的终结似乎遥不可及。许多人在编程 VCR 时遇到的困难是这个问题的一个很小但真实的例子：让计算机做用户想让它做的事情仍然像以往一样困难，需要天赋、辛勤工作以及开发人员对用户需求的承诺。

By the mid-1990s personal computers had become a commodity, allowing commercial software to come to the fore as the central place where innovation was conveyed to users. The layering of software, a process that began with the first “automatic coding” schemes developed for the UNIVAC, continued. That was the only way to broaden the market to include users who had no inclination or talent to write programs. Again, with each software advance, one heard that the “end of programming” had come, that “anyone” could now get a computer to do what he or she wished. As new markets opened up, the end proved elusive. The difficulty many people have in programming a VCR is a minor but real example of the problem: getting a computer to do what users want it to do is as difficult as ever and requires talent, hard work and a commitment by developers to the user’s needs.

Macintosh 界面为个人电脑带来了易用性，而微软的 Windows 也效仿了这种易用性，这又带来了一系列新的挫折。用户现在发现，这些界面本应使计算变得更加容易，但事实上，它们让事情变得更加困难。毫无疑问，这一过程将继续下去。2001 年已经过去，但智能计算机 HAL 却并未问世，HAL 就是斯坦利·库布里克的电影《2001 太空漫游》中的主角。很多人看完电影后，认为 HAL 的问题在于它不知何故失去了控制；但仔细观察就会发现，HAL 的真正问题在于它运行良好。它之所以崩溃，是因为它试图遵循编程中的两个相互冲突的指令：服从机上的人类，但向他们隐瞒任务的真实性质。1^如果真的出现了类似 HAL 的智能界面，那么它可能不会像虚构的界面那样强大和可靠。

The ease of use that the Macintosh interface brought to personal computing, which Microsoft copied with Windows, has led to a new set of frustrations. Users now find interfaces laid over these interfaces, which are supposed to make computing even easier. In fact, they have made things more difficult. This process will doubtless continue. The year 2001 has come and gone, and it did not bring with it a realization of the intelligent computer HAL, the star of Stanley Kubrick’s movie 2001 A Space Odyssey. Many people came away from the movie thinking that the problem with HAL was that it was somehow out of control; but a closer viewing shows that HAL’s real problem was that it worked perfectly. It broke down because it was trying to obey two conflicting instructions that were part of its programming: to obey the humans on board but to conceal from them the true nature of their mission.¹ If a real version of a HAL-like intelligent interface ever appears, it will probably not be as robust and reliable as the fictional one.

世界图景的数字化

The Digitization of the World Picture

1948 年，一本名为《世界图景的机械化》的书问世。作者是荷兰物理学家 EJ Dijksterhuis，他认为，历史的大部分内容最好理解为“机械”的世界观的展开，这种世界观实际上始于希腊人，并在艾萨克·牛顿的著作中达到顶峰。2 ^Dijksterhuis的作品吸引了一批热心读者，他们在六年的世界大战之后体验了机械化世界观的力量和恐怖。

In 1948 a book appeared with the bold title The Mechanization of the World Picture. The author, a Dutch physicist named E. J. Dijksterhuis, argued that much of history was best understood as an unfolding of the “mechanistic” way of looking at the world that actually began with the Greeks and culminated in the work of Isaac Newton.² Dijksterhuis’s work found a willing audience of readers who had experienced the power and the horrors of a mechanized world view after six years of world war.

机械论的观点花了一千五百年才站稳脚跟，但同样具有革命性的观点却只用了更短的时间——大约五十年——就站稳了脚跟。“世界图景的数字化”始于 20 世纪 30 年代中期，由几位数学家和工程师发起。到 1985 年，这种世界观取得了胜利。它始于数学中一个不起眼的角落。阿兰·图灵在 1936 年的一篇论文中提出的“机器”是一种理论构造。3^存储程序电子计算机的发明使他的想法焕发生机，并使其比他想象的更真实。在随后的几十年里，计算机就像一种万能溶剂一样，接管、吸收或改造了一个又一个领域。4 1973 年 10 月，行业杂志《电子学》的一期特刊将^传统模拟电子电路被通过编程模拟它们的微型数字计算机所取代的过程描述为“大接管”；例如，到 1973 年，大多数普通收音机都不再有调谐拨盘，而是通过数字键盘进行“调谐”。十年后，《时代》杂志^将计算机评为1983 年“年度机器”，开篇标题是“计算机进军”。5

It took a millennium and a half for a mechanistic view to take hold, but it has taken less time—about fifty years—for a view equally as revolutionary to take hold. The “digitization of the world picture” began in the mid-1930s, with the work of a few mathematicians and engineers. By 1985 this world view had triumphed. It began in an obscure corner of mathematics. Alan Turing’s “machine,” introduced in a paper in 1936, was a theoretical construction.³ The invention of the stored-program electronic computer breathed life into his idea and made it more real than he probably thought possible. The ensuing decades saw one field after another taken over, absorbed, or transformed by the computer as if it were a universal solvent.⁴ A special issue of the trade journal Electronics, in October 1973 described as “The Great Takeover,” the way traditional analog electronic circuits were replaced by miniature digital computers programmed to emulate them; most ordinary radios, for example, had lost their tuning dial by 1973 and were “tuned” by digital keypads. Ten years later, Time proclaimed the computer “Machine of the Year” for 1983, with the opening headline “The Computer Moves In.”⁵

这种接管的最新表现是互联网，受到纽特·金里奇、阿尔·戈尔、斯图尔特·布兰德、已故的蒂莫西·利里、“X 一代”以及众多介于两者之间的人士的广泛欢迎。大多数报道将其描述为通信和计算的结合。6这里提供的证据表明情况并非^如此；互联网只是代表了另一次接管，数字计算对一项长期以来基于模拟技术的活动（电信）进行了接管。

The latest manifestation of this takeover is the Internet, embraced across the political and cultural spectrum, by Newt Gingrich, Al Gore, Stewart Brand, the late Timothy Leary, “Generation X,” and numerous people in between. Most accounts describe it as a marriage of communications and computing.⁶ The evidence presented here suggests otherwise; that the Internet simply represents yet another takeover, by digital computing of an activity (telecommunications) that had a long history based on analog techniques.

那些热情洋溢地将万维网描述为 50 年序幕的顶峰的人要么不了解历史，要么忘记了历史。当第一台 UNIVAC 安装时，当小型计算机和分时出现时，当个人计算机推出时，都曾有过同样的说法（图 C.1）。这不会是他们最后一次说这些话。但是，在美国历史上，对技术乌托邦的承诺并不罕见，至少有一些互联网拥护者知道这些早期的愿景有多么天真。7^硅谷拥有全美最拥挤的真实高速公路之一，因为人们上下班时使用亨利·福特为缓解城市交通拥堵而发明的技术。大多数人都知道，汽车之所以未能兑现福特的许多承诺，仅仅是因为它太成功了。“烟雾”一词在福特于 20 世纪 40 年代末去世前后悄悄进入了英语； “交通堵塞”、“商业街”和“郊区扩张”等词汇后来才出现。什么词汇能描述网络数字计算的阴暗面？这些“副作用”是否会像汽车一样在 50 年后才显现出来？我们能否在为时已晚或难以管理之前预见到它们？

Those who so glowingly describe the World Wide Web as the culmination of fifty years of prologue either do not know or have forgotten history. The very same statements were made when the first UNIVACs were installed, when minicomputers and time-sharing appeared, and when the personal computer was introduced (figure C.1). This will not be the last time these words are spoken. But promises of a technological Utopia have been common in American history, and at least a few champions of the Internet are aware of how naive these earlier visions were. ⁷ Silicon Valley has some of the most congested real highways in the country, as people commute to work with a technology that Henry Ford invented to reduce urban congestion. Most people have some sense of the fact that the automobile did not fulfill many of Ford’s promises simply because it was too successful. The word “smog” crept into the English language around the time of Ford’s death in the late 1940s; “gridlock,” “strip malls,” and “suburban sprawl” came later. What equivalent will describe the dark side of networked digital computing? And will those “side effects” become evident only fifty years from now, as was the case with automobiles? Can we anticipate them before it is too late or too difficult to manage them?

每次数字计算的变革都是由抱有理想主义观念的人推动的，他们认为新形式的计算将成为一股解放力量，可以纠正阿尔文·托夫勒所说的“第二次浪潮”烟囱带来的许多不平衡。UNIVAC 的安装伴随着热情洋溢的预测，即它们产生的“自动化”将减少每周的工作时间。20 世纪 60 年代中期，爱好者和黑客将 PDP-10 和 PDP-8 视为将计算从 IBM 章鱼的触角中解放出来的机器。Apple II 反映了 20 世纪 70 年代初旧金山湾区的乌托邦愿景。互联网的普遍接入也将如此。

Each transformation of digital computing was propelled by individuals with an idealistic notion that computing, in its new form, would be a liberating force that could redress many of the imbalances brought on by the smokestack of the “second wave,” in Alvin Toffler’s phrase. UNIVAC installations were accompanied by glowing predictions that the “automation” they produced would lead to a reduced workweek. In the mid-1960s enthusiasts and hackers saw the PDP-10 and PDP-8 as machines that would liberate computing from the tentacles of the IBM octopus. The Apple II reflected the Utopian visions of the San Francisco Bay area in the early 1970s. And so it will be with universal access to the Internet.

图 C.1数字乌托邦，如Byte

杂志封面所示（1977 年 1 月）。Byte的封面插图在所有计算机出版物中脱颖而出。（来源：罗伯特·廷尼。）

Figure C.1

Digital Utopia, as depicted on the cover of Byte magazine (January 1977). Byte’s cover illustrations stood out among all the computer publications. (Source : Robert Tinney.)

在每种情况下，未来都比其倡导者想象的更加复杂，也更缺乏革命性。UNIVAC 未能解决失业问题。个人电脑未能让普通人与掌权者平等。它确实找到了一个超出所有预期的市场——但在办公室而不是家庭，作为协助企业工作场所功能的工具。8 20 世纪 70 年代，年轻人看着工业锈带污染和腐朽的景象，对他们的个人电脑进行编程，以模拟一个中间景观；一个让居民享受工业化所有好处而没有任何缺点的景观。但外部世界的社会^问题仍然存在。乌托邦留在电脑屏幕内，顽固地拒绝出来。计算机建模演变成“虚拟现实”——20 世纪 60 年代流行的精神改变药物的新变种。蒂莫西·利里认为，虚拟现实比 LSD 更能有效地将人类带回伊甸园。到目前为止，这种情况还没有发生，但考虑到大多数人对“数字乌托邦”的设想的思考水平，这或许是一件好事。

In each case the future has turned out to be more complex, and less revolutionary, than its proponents imagined. The UNIVAC did not solve the problem of unemployment. Personal computers did not put ordinary individuals on an equal footing with those in positions of power. It did find a market that exceeded all expectations—but in the office and not the home, as a tool that assisted the functions of the corporate workplace.⁸ Looking out over the polluted and decayed landscape of the 1970s-era industrial Rustbelt, young people programmed their personal computers to model a middle landscape; one that gave its inhabitants all the benefits of industrialization with none of the drawbacks. But the social problems of the outside world remained. Utopia stayed inside the computer screen and stubbornly refused to come out. Computer modeling evolved into “virtual reality”—a new variant of the mind-altering drugs in vogue in the 1960s. Timothy Leary argued that virtual reality was more effective than LSD as a way to bring humans back to the Garden of Eden. So far that is not happening, and perhaps this is a good thing, given the level of thought that characterizes most visions of what Digital Utopia ought to look like.

我们已经看到，政治和社会力量始终在塑造数字计算的发展方向。现在，随着计算成为美国社会的决定性技术之一，这些力量越来越公开化，成为公众讨论的一部分。政客和法官与工程师一样，决定在哪里修建高速公路和桥梁，谁可以为某个地区提供电话服务，以及电力公司可能面临多少竞争。这些立法者和法学家依靠行业游说者或员工中的专家来指导他们制定政策的技术层面。与此同时，新技术（如直接广播卫星电视）扰乱了他们的计划。但这并没有阻止这一进程，也没有将决策权从这些中心转移出去。

We have seen that political and social forces have always shaped the direction of digital computing. Now, with computing among the defining technologies of American society, those forces are increasingly out in the open and part of public discussion. Politicians and judges as much as engineers decide where highways and bridges get built, who may serve a region with telephone service, and how much competition an electric utility may have. These legislators and jurists rely upon industry lobbyists or specialists on their staff to guide them through the technical dimension of their policies. All the while, new technologies (such as direct broadcast satellite television) disrupt their plans. But that does not stop the process or shift decision-making away from these centers.

计算机也不例外。政治家指导技术的想法仍然令计算机先驱们反感，他们中的许多人仍然活着，并清楚地记得他们如何克服技术挑战，而不是政治挑战。但当一项技术融入日常生活时，它必须承认政治。一些组织，如电子前沿基金会（由米奇·卡普尔创立），正在通过退一步来尝试识别“雾霾”和“交通堵塞”的数字对应物。但从历史上看，美国

Computing is no different. The idea of politicians directing technology is still distasteful to computer pioneers, many of whom are still alive and retain a vivid memory of how they surmounted technical, not political, challenges. But when a technology becomes integrated into the affairs of ordinary daily life, it must acknowledge politics. Some groups, such as the Electronic Frontier Foundation (founded by Mitch Kapor), are doing this by stepping back to try to identify the digital equivalents of “smog” and “gridlock.” But historically the United States

推动了技术尽可能快速的部署，并将后果留给后代来处理。因此，迄今为止，试图监管或控制互联网内容的努力笨拙而失败，这并不奇怪。结果如何还有待观察。

has promoted as rapid a deployment of technology as possible, and has left it to future generations to deal with the consequences. It is not surprising, therefore, that attempts to regulate or control the content of the Internet have so far been clumsy and have failed. How that plays out remains to be seen.

一个半世纪前，亨利·戴维·梭罗曾怀疑地观察到美国人对技术的痴迷。铁路是他那个时代的高科技，但他并不认同公众对菲奇堡铁路的热情，该铁路的轨道位于瓦尔登湖后面。“我们不是搭乘铁路，而是铁路搭乘我们，”他说。国家需要的是“一种严肃而又超越斯巴达式的简单生活”。在梭罗小屋以西几英里处，菲奇堡铁路修建了一条支线，为阿萨贝特磨坊提供服务，该磨坊在南北战争时期是美国最大的羊毛制品生产商之一。一个世纪后，这些磨坊的 PDP-8 遍布全球。人们不禁想知道梭罗会如何看待这种联系。9^他会抓住机会建立自己的瓦尔登湖主页，让其他人知道他在做什么吗？还是会继续依赖自己制作的铅笔？

A century and a half ago, Henry David Thoreau observed with suspicion the technophilic aspect of American character. Railroads were the high technology of his day, but he did not share the public’s enthusiasm for the Fitchburg line, whose tracks ran behind Walden Pond. “We do not ride on the railroad; it rides on us,” he said. What the nation needs is “a stern and more than Spartan simplicity of life.” A few miles west of Thoreau’s cabin, the Fitchburg railroad built a branch to serve the Assabet Mills, which by the time of the Civil War was one of the country’s largest producers of woolen goods. A century later these same mills were blanketing the Earth with PDP-8s. One wonders what Thoreau would have made of this connection.⁹ Would he have seized the opportunity to set up his own Walden Pond home page, to let others know what he was up to? Or would he have continued to rely on the pencils he made for himself?

我们创造了计算机来为我们服务。几十年来，计算机可能成为我们的主人这一想法一直是科幻小说的题材，但当让计算机做基本的家务需要付出巨大的努力时，我们很难认真对待这些故事。当我们开始接受万维网作为我们日常生活的自然组成部分时，也许是时候重新审视控制问题了。我希望，通过了解历史和一点梭罗式的怀疑论，我们可以学会使用计算机，而不是让它使用我们。

We created the computer to serve us. The notion that it might become our master has been the stuff of science fiction for decades, but it was always hard to take those stories seriously when it took heroic efforts just to get a computer to do basic chores. As we start to accept the World Wide Web as a natural part of our daily existence, perhaps it is time to revisit the question of control. My hope is that, with an understanding of history and a dash of Thoreauvian skepticism, we can learn to use the computer rather than allowing it to use us.

笔记

Notes

前言

Preface

1. Jim Clark 与 Owen Edwards 合著，《Netscape Time：挑战微软的十亿美元新兴公司的诞生》（纽约：圣马丁出版社，1999 年）。

1. Jim Clark, with Owen Edwards, Netscape Time: The Making of the Billion-Dollar Start-Up That Took on Microsoft (New York: St. Martin’s Press, 1999).

介绍

Introduction

1. 所有的词典，甚至是 20 世纪 70 年代出版的版本，都将计算机定义为“专为解决复杂数学问题而设计的计算器”。这是《韦氏第三版国际词典》未删节版中给出的第一个定义；该定义被限定为“具体而言：一种可以存储、检索和处理数据的可编程电子设备”。

1. Dictionaries, even editions published in the 1970s, define computer as “a calculator especially designed for the solution of complex mathematical problems.” This is the first definition given in Webster’s Third International Dictionary, Unabridged; this definition is then qualified as “specifically: a programmable electronic device that can store, retrieve, and process data.”

2.一些早期的自动机器被称为“计算器”，例如 Harvard Mark I，即“自动序列控制计算器”。但 ENIAC 中的字母“C”代表“Computer”（计算机），它是在 20 世纪 40 年代初由摩尔学院设计并于 1946 年投入使用的。

2. Some of the early automatic machines were called “calculators,” as in the Harvard Mark I, or “Automatic Sequence Controlled Calculator.” But the letter “C” in ENIAC, designed at the Moore School in the early 1940s and dedicated in 1946, stood for “Computer.”

3. Amy Friedlander，《自然垄断与普遍服务： 1837-1940 年美国通信基础设施中的电话和电报》（弗吉尼亚州雷斯顿：CNRI，1995 年）。

3. Amy Friedlander, Natural Monopoly and Universal Service: Telephones and Telegraphs in the U.S. Communications Infrastructure, 1837–1940 (Reston, VA: CNRI, 1995).

4.无论如何，情况可能恰恰相反：技术史学家正将注意力转向平凡的事物；而计算研究如此普遍，以至于没有人会感到惊讶。例如，请参阅亨利·佩特罗斯基的《铅笔：设计和环境的历史》（纽约：克诺夫出版社，1990 年）和罗伯特·弗里德尔的《拉链：新奇探索》（纽约：WW·诺顿出版社，1994 年）。

4. If anything, it might go the other way: historians of technology are turning their attention to the mundane; and studies of computing are so common they surprise no one. See, for example, Henry Petroski, The Pencil: a History of Design and Circumstance (New York: Knopf, 1990), and Robert Friedel, Zipper: an Exploration in Novelty (New York: W.W. Norton, 1994).

5.例如，请参阅 I. Bernard Cohen 著《科学革命》（剑桥：哈佛大学出版社，1985 年）。

5. See, for example I. Bernard Cohen, Revolution in Science (Cambridge: Harvard University Press, 1985).

6.当我开始这项研究时，我还没有听说过万维网，尽管我知道互联网的存在。尽管《时代》杂志在1983 年的封面上将个人电脑吹捧为革命性的，但现在它却因其粗糙的用户界面和缺乏网络连接而受到贬低，被认为存在缺陷。

6. I had not heard of the World Wide Web when I began working on this study, although I was aware of the existence of the Internet. Although touted as revolutionary by Time on its cover in 1983, the personal computer is now disparaged as crippled by its crude user interface and lack of connectivity to the Web.

7.我这一观察的非科学依据是学者们在万维网上对技术史的积极研究。我还注意到，历史学家是第一批采用最新文字处理和学者数据库工具的人之一。

7. My unscientific basis for this observation is the vigorous activity in the history of technology being undertaken by scholars on the World Wide Web. I have also notestext1d that historians are among the first to adopt the latest word processing and scholars’ database tools.

8.例如，请参阅阿尔文·托夫勒的《第三次浪潮》（纽约：莫罗，1980 年）。

8. See, for example, Alvin Toffler, The Third Wave (New York: Morrow, 1980).

9.即使是最好的计算机社会学研究也忽视了它的历史演变，好像社会学家观察到的技术是既定事实，而不是某种快速发展的事物；例如，Shoshanna Zuboff 的《智能机器时代》（纽约：Basic Books，1988 年）和 Sherry Turkle 的《第二自我、计算机和人类精神》（纽约：西蒙与舒斯特出版社，1984 年）。

9. Even the best sociological studies of computing ignore its historical evolution, as if the technology sociologists observe is a given and not something that is rapidly evolving; for example, Shoshanna Zuboff, In the Age of the Smart Machine (New York: Basic Books, 1988), and Sherry Turkle, The Second Self, Computers and the Human Spirit (New York: Simon & Schuster, 1984).

10.例如，请参阅 James W. Cortada 的《计算机出现之前》（普林斯顿：普林斯顿大学出版社，1993 年）；Arthur Norberg 的《20 世纪早期的高科技计算：商业和政府中的打孔卡机器》，《科技与文化》第 31 卷（1990 年 10 月）：753-779 页；以及 JoAnne Yates 的《通过沟通进行控制：美国管理系统的兴起》（巴尔的摩：约翰霍普金斯大学出版社，1989 年）。

10. See, for example, James W. Cortada, Before the Computer (Princeton: Princeton University Press, 1993); Arthur Norberg, “High Technology Calculation in the Early Twentieth Century: Punched Card Machinery in Business and Government,” Technology and Culture 31 (October 1990): 753–779; and JoAnne Yates, Control Through Communication: the Rise of System in American Management (Baltimore: John Hopkins University Press, 1989).

11.例如，请参阅 James R. Beniger 著《控制革命：信息社会的技术和经济起源》（剑桥：哈佛大学出版社，1986 年）。

11. See, for example, James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society (Cambridge: Harvard University Press, 1986).

12.例如，《计算机历史年鉴》是研究这一主题的权威期刊，它很少发表将计算机与雷达、弹道导弹或核武器历史联系起来的论文，除非是讨论计算机作为这些技术的助手所起的作用。另一方面，人们发现 20 世纪现代技术的历史完全没有提到计算机，例如托马斯·帕克·休斯的《美国起源：一个世纪的发明和技术热情，1870-1970》（纽约：维京出版社，1989 年）。

12. For example, the Annals of the History of Computing, the journal of record for the study of this subject, seldom publishes papers that connect computing with, say, radar, ballistic missiles, or nuclear weapons history, other than on the role of the computer as an aide to those technologies. On the other side, one finds histories of modern 20th century technology that make no mention of the computer at all, as in Thomas Parke Hughes, American Genesis: A Century of Invention and Technological Enthusiasm, 1870–1970 (New York: Viking, 1989).

13. Thomas Parke Hughes，《电力网络：1880-1930 年西方社会的电气化》（巴尔的摩：约翰霍普金斯大学出版社，1983 年）。

13. Thomas Parke Hughes, Networks of Power: Electrification in Western Society, 1880–1930 (Baltimore: Johns Hopkins University Press, 1983).

14.关于这个话题的众多著作中，最容易理解的一本是 Wiebe E. Bijker、Thomas P. Hughes 和 Trevor Pinch 编的《技术系统的社会建构》（剑桥：麻省理工学院出版社，1987 年）。

14. The most accessible of the many works written on this topic is Wiebe E. Bijker, Thomas P. Hughes, and Trevor Pinch, eds., The Social Construction of Technological Systems (Cambridge: MIT Press, 1987).

15.最重要的是唐纳德·麦肯齐。例如，参见《发明准确性：核导弹制导的历史社会学》（剑桥：麻省理工学院出版社，1990 年）和《协商算术，构建证明：数学和信息技术的社会学》，《科学社会研究》第 23 卷（1993 年）：第 37-65 页。另一位从业者是布莱恩·普法芬伯格；参见他的《个人电脑的社会意义，或为什么个人电脑革命不是革命》，《人类学季刊》第 61 卷：第 1 期（1988 年）：第 39-47 页。

15. The most important is Donald MacKenzie. See, for example, Inventing Accuracy: A Historical Sociology of Nuclear Missile Guidance (Cambridge: MIT Press, 1990), and “Negotiating Arithmetic, Constructing Proof: the Sociology of Mathematics and Information Technology,” Social Studies of Science 23 (1993): 37–65. Another practitioner is Bryan Pfaffenberger; see his “The Social Meaning of the Personal Computer, or Why the Personal Computer Revolution was no Revolution,” Anthropological Quarterly 61: 1 (1988): 39–47.

16.例如，请参阅 Steven Levy 的《黑客：计算机革命的英雄》（纽约 Garden City：Doubleday，1984 年）；以及 Paul Freiberger 的《硅谷之火：个人电脑的诞生》（伯克利：Osborne=McGraw-Hill，1984 年）。

16. See, for example Steven Levy, Hackers: Heroes of the Computer Revolution (Garden City, NY: Doubleday, 1984); and Paul Freiberger, Fire in the Valley: the Making of the Personal Computer (Berkeley: Osborne=McGraw-Hill, 1984).

17. William Aspray，《约翰·冯·诺依曼与现代计算的起源》（剑桥：麻省理工学院出版社，1990 年）。

17. William Aspray, John von Neumann and the Origins of Modern Computing (Cambridge: MIT Press, 1990).

18. Tomas J. Misa，“军事需求、商业现实和晶体管的发展，1948-1958”，载 Merritt Roe Smith 主编，《军事企业和技术变革》（剑桥：麻省理工学院出版社，1985 年）：第 253-287 页。

18. Tomas J. Misa, “Military Needs, Commercial Realities, and the Development of the Transistor, 1948–1958,” in Merritt Roe Smith, ed., Military Enterprise and Technological Change (Cambridge: MIT Press, 1985): 253–287.

19. Michael A. Dennis，“国家变迁：1930-1945 年麻省理工学院仪器实验室和约翰霍普金斯大学应用物理实验室技术实践的政治文化”（博士论文，约翰霍普金斯大学，1990 年）。

19. Michael A. Dennis, “A Change of State: the Political Cultures of Technological Practice at the MIT Instrumentation Lab and the Johns Hopkins University Applied Physics Laboratory, 1930–1945” (Ph.D. diss., Johns Hopkins University, 1990).

20. Manuel DeLanda，《智能机器时代的战争》（剑桥：麻省理工学院出版社，1991 年）；另请参阅 Chris Hables Gray 的《计算机作为武器和隐喻：1940-1990 年的美国军队和后现代战争》（博士论文，加利福尼亚大学，圣克鲁斯分校，1991 年）。

20. Manuel DeLanda, War in the Age of Intelligent Machines (Cambridge: MIT Press, 1991); also Chris Hables Gray, “Computers as Weapons and Metaphors: The U.S. Military 1940–1990 and Postmodern War,” (Ph.D. diss., University of California, Santa Cruz, 1991).

21. Charles Bashe、Lyle R. Johnson、John H. Palmer 和 Emerson Pugh，《IBM 的早期计算机》（剑桥：麻省理工学院出版社，1986 年）；Emerson Pugh 等，《IBM 的 360 和早期 370 系统》（剑桥：麻省理工学院出版社，1991 年）；以及 Emerson Pugh，《打造 IBM：塑造一个行业及其技术》（剑桥：麻省理工学院出版社，1995 年）。

21. Charles Bashe, Lyle R. Johnson, John H. Palmer, and Emerson Pugh, IBM’s Early Computers (Cambridge: MIT Press, 1986); Emerson Pugh et al., IBM’s 360 and Early 370 Systems (Cambridge: MIT Press, 1991); and Emerson Pugh, Building IBM: Shaping an Industry and Its Technology (Cambridge: MIT Press, 1995).

22.该术语似乎在 1959 年左右开始使用。

22. The term seems to have come into use around 1959.

23. George H. Mealy，《操作系统》，载于 Saul Rosen 主编的《编程系统和语言》（纽约：McGraw-Hill，1967 年）：517-518。

23. George H. Mealy, “Operating Systems,” in Saul Rosen, ed., Programming Systems and Languages (New York: McGraw-Hill, 1967): 517–518.

24. JoAnne Yates，《通过沟通进行控制：美国管理体制的兴起》（巴尔的摩：约翰霍普金斯大学出版社，1989年）；David F. Noble，《生产力》（纽约：克诺夫出版社，1984年）；James R. Beniger，《控制革命：信息社会的技术和经济起源》（剑桥：哈佛大学出版社，1986年）。

24. JoAnne Yates, Control Through Communication: the Rise of System in American Management (Baltimore: Johns Hopkins University Press, 1989); David F. Noble, Forces of Production (New York: Knopf, 1984); James R. Beniger, The Control Revolution: Technological and Economic Origins of the Information Society (Cambridge: Harvard University Press, 1986).

25. Brian Randell 编，《数字计算机的起源：精选论文》，第 2 版。（柏林：Springer-Verlag，1975 年）：327-328；Peter J. Bird，《LEO：第一台商用计算机》（英国伯克郡：Hasler Publishing Ltd.，1994 年）。

25. Brian Randell, ed., The Origins of Digital Computers: Selected Papers, 2nd ed. (Berlin: Springer-Verlag, 1975): 327–328; Peter J. Bird, LEO: the First Business Computer (Berkshire, UK: Hasler Publishing Ltd., 1994).

26. Kenneth Flamm，《创造计算机：政府、工业和高科技》（华盛顿特区：布鲁金斯学会，1988 年）：134；另请参阅 Martin Campbell-Kelly，《ICL：商业和技术史》（牛津：牛津大学出版社，1989 年）。

26. Kenneth Flamm, Creating the Computer: Government, Industry, and High Technology (Washington, DC: Brookings Institution, 1988): 134; see also Martin Campbell-Kelly, ICL: a Business and Technical History (Oxford: Oxford University Press, 1989).

27. Edward Feigenbaum 和 Pamela McCorduck，《第五代：人工智能和日本计算机对世界的挑战》（马萨诸塞州雷丁：Addison-Wesley，1983 年）；Michael Cusumano，“软件开发中的工厂概念和实践”，《计算机史年鉴》 13：1（1991 年）：3-32。

27. Edward Feigenbaum and Pamela McCorduck, The Fifth Generation: Artificial Intelligence and Japan’s Computer Challenge to the World (Reading, MA: Addison-Wesley, 1983); Michael Cusumano, “Factory Concepts and Practices in Software Development,” Annals of the History of Computing 13: 1 (1991): 3–32.

28. Seymour E. Goodman，“苏联计算和技术转让：概述”，世界政治31:4（1979 年 7 月）：539-570。

28. Seymour E. Goodman, “Soviet Computing and Technology Transfer: an Overview,” World Politics 31: 4 (July 1979): 539–570.

29.赫尔曼·卡恩和 R. 巴克敏斯特·富勒等美国“未来学家”曾大力推崇利用计算机进行集中规划。在完成旋风计划后，J. 福雷斯特转向了一种他称之为“系统动力学”的应用程序。1996 年，美国政府的大规模计算机建模得到了总统候选人、电子数据系统公司创始人 H. 罗斯·佩罗的大力推动。

29. Using the computer for centralized planning has been touted by American “futurists” such as Herman Kahn and R. Buckminster Fuller. After completing Project Whirlwind, J. Forrester turned to an application he called “System Dynamics.” In 1996, large-scale computer modeling of the U.S. government was vigorously promoted by presidential candidate H. Ross Perot, the founder of Electronic Data Systems.

第1章

Chapter 1

1.坎农的证词，哈格利博物馆，霍尼韦尔诉斯佩里兰德案卷，系列 III，第 140 盒，第 17,680 页；另见 Harold Bergstein 的《对 Eckert 和 Mauchly 的采访》，Datamation（1962 年 4 月）：25-30。I. Bernard Cohen 即将出版的一本关于霍华德·艾肯生平和作品的书将对艾肯的观察进行更详细的分析。我非常感谢 Cohen 教授在本书出版前向我提供了相关章节的草稿。

1. Testimony by Cannon, Hagley Museum, Honeywell v. Sperry Rand papers, Series III, Box 140, p. 17,680; see also Harold Bergstein, “An Interview with Eckert and Mauchly,” Datamation (April 1962): 25–30. A more detailed analysis of Aiken’s observation is discussed in a forthcoming book by I. Bernard Cohen on the life and work of Howard Aiken. I am grateful to Professor Cohen for making drafts of the relevant chapters of this book available to me before its publication.

2. notestext1 指出，1994 年美国政府暂停了对超导超级对撞机 (SSC) 的支持。因此，从中可以看出，全球“市场”总量已达到顶峰，大约有十几台回旋加速器，这是一种与电子计算机同时发明的科学仪器，其成本和复杂性大致相同。

2. notestext1 that in 1994 the U.S. government suspended support for the Superconducting Super Collider (SSC). So it appears there that the total world “market” has peaked at about a dozen cyclotrons, a scientific instrument invented around the same time as the electronic computer with about the same cost and complexity.

3.例如，冯·诺依曼在 1946 年 5 月对美国海军数学计算顾问小组的演讲中，将当时正在开发的电子计算机比作“……目前仍然是主要实用的计算模式，即使用机台式乘法器的人工操作”。发表于《计算机史年鉴》第 10 卷（1989 年）：第 248 页。

3. For example, in an address by von Neumann to the Mathematical Computing Advisory Panel of the U.S. Navy in May 1946, he compares the electronic computers then under development to “ ... what is at present still the major practical mode of computing, namely, human procedure with an electromechanical desk multiplier.” Published in the Annals of the History of Computing 10 (1989): 248.

4.有关此项活动早期发展的叙述，请参阅 JoAnne Yates 的《通过沟通进行控制：美国管理体制的兴起》（巴尔的摩：约翰霍普金斯大学出版社，1989 年）；另请参阅 James Beniger 的《控制革命》（剑桥：哈佛大学出版社，1986 年）。

4. For an account of the early development of this activity, see JoAnne Yates, Control Through Communication: the Rise of System in American Management (Baltimore: Johns Hopkins University Press, 1989); also James Beniger, The Control Revolution (Cambridge: Harvard University Press, 1986).

5. Arthur Norberg，《20 世纪早期的高科技计算：商业与政府中的打孔卡机器》，《科技与文化》第 31 卷（1990 年）：753-779 页；另请参阅 Martin Campbell-Kelly 所著《ICL：商业与技术史》（牛津：牛津大学出版社，1989 年）。

5. Arthur Norberg, “High Technology Calculation in the Early Twentieth Century: Punched Card Machinery in Business and Government,” Technology and Culture 31 (1990): 753–779; also Martin Campbell-Kelly, ICL: a Business and Technical History (Oxford: Oxford University Press, 1989).

6.以下有关打孔卡计算的讨论源自 Martin Campbell-Kelly 所著《打孔卡机械》，收录于 William Aspray 主编的《计算机之前的计算》（艾姆斯：爱荷华州立大学出版社，1990 年），第 4 章；另请参阅 Campbell-Kelly 所著的《ICL》；以及 Edmund C. Berkeley 所著的《巨大的大脑，或会思考的机器》（纽约：威利出版社，1949 年），第 4 章。

6. The following discussion on punched-card computation is derived from Martin Campbell-Kelly, “Punched-Card Machinery,” in William Aspray, ed., Computing Before Computers (Ames: Iowa State University Press, 1990), chapter 4; also Campbell-Kelly, ICL; and Edmund C. Berkeley, Giant Brains, or Machines that Think (New York: Wiley, 1949), chapter 4.

7. Campbell-Kelly，在 Aspray 编辑的《计算机之前的计算》中；此外还有Kenneth Flamm 在《计算史年鉴》 13: 1 (1991)中对 Campbell-Kelly 的ICL的评论。

7. Campbell-Kelly, in Aspray, ed., Computing Before Computers; also the review of Campbell-Kelly’s ICL by Kenneth Flamm in Annals of the History of Computing 13: 1 (1991).

8. Wallace J. Eckert，《科学计算中的穿孔卡方法》（纽约：哥伦比亚大学 Thomas J. Watson 天文计算局，1940 年）：22。

8. Wallace J. Eckert, Punched Card Methods in Scientific Computation (New York: The Thomas J. Watson Astronomical Computing Bureau, Columbia University, 1940): 22.

9.同上，第1页。

9. Ibid., 1.

10.同上，第 108–110 页。十二个步骤中，只有前六个步骤是自动执行的，其余步骤则需要人工干预。

10. Ibid., 108–110. Of the twelve steps, only the first six were performed automatically; the rest required some human intervention.

11. J. Lynch 和 CE Johnson，《IBM 中继计算器的编程原理》，弹道研究实验室，报告编号 705，1949 年 10 月，IBM 档案馆，纽约 Valhalla。

11. J. Lynch and C. E. Johnson, “Programming Principles for the IBM Relay Calculators,” Ballistic Research Laboratories, Report No. 705, October 1949, IBM Archives, Valhalla, New York.

12.同上，第8页。

12. Ibid., 8.

13. Brian Randell 编，《数字计算机的起源：精选论文》，第二版（柏林：Springer-Verlag，1975 年）：188。

13. Brian Randell, ed., The Origins of Digital Computers: Selected Papers, 2d ed. (Berlin: Springer-Verlag, 1975): 188.

14. Lynch 和 Johnson，《编程原理》，第 4 页；另请参阅 Wallace Eckert，《IBM 可插拔序列中继计算器》，《数学表格和其他计算辅助工具》第 3 卷 (1948)：第 149-161 页；另请参阅弹道研究实验室，《计算实验室》，未注明日期的 15 页手册，可能是 1952 年出版，国家航空航天博物馆，NBS 收藏。

14. Lynch and Johnson, “Programming Principles,” 4; also Wallace Eckert, “The IBM Pluggable Sequence Relay Calculator,” Mathematical Tables and Other Aids to Computation 3 (1948): 149–161; also Ballistic Research Laboratories, “Computing Laboratory,” undated 15-page brochure, probably 1952, National Air and Space Museum, NBS Collection.

15. Charles J. Bashe、Lyle R. Johnson、John H. Palmer 和 Emerson Pugh，《IBM 的早期计算机》（剑桥：麻省理工学院出版社，1986 年）：44–46、59–68。

15. Charles J. Bashe, Lyle R. Johnson, John H. Palmer, and Emerson Pugh, IBM’s Early Computers (Cambridge: MIT Press, 1986): 44–46, 59–68.

16.同上，第67页。

16. Ibid., 67.

17. William Woodbury，《603-405 计算机》，载于《第二次计算机研讨会论文集；1949 年 9 月》（剑桥：哈佛大学出版社，1951 年）：第 316-320 页；另载于 Michael R. Williams，《计算技术史》（新泽西州 Englewood Cliffs：Prentice-Hall，1985 年）：第 256 页。

17. William Woodbury, “The 603-405 Computer,” in Proceedings of a Second Symposium on Calculating Machinery; Sept. 1949 (Cambridge: Harvard University Press, 1951): 316–320; also Michael R. Williams, A History of Computing Technology (Englewood Cliffs, NJ: Prentice-Hall, 1985): 256.

18. GJ Toben，引自Bashe等人所著《IBM的早期计算机》，第69页。

18. G. J. Toben, quoted in Bashe et al., IBM’s Early Computers, 69.

19. Bashe 等人，《IBM 的早期计算机》，第 68-72 页；此外，John W. Sheldon 和 Liston Tatum 的《IBM 卡片编程电子计算器》，《电子数字计算机评论》，IRE-AIEE 联合会议，1952 年 2 月，第 30-36 页。

19. Bashe et al., IBM’s Early Computers, 68–72; also John W. Sheldon and Liston Tatum, “The IBM Card-Programmed Electronic Calculator,” Review of Electronic Digital Computers, Joint IRE-AIEE Conference, February 1952, 30–36.

20. Paul Ceruzzi，《超越极限：飞行进入计算机时代》（剑桥：麻省理工学院出版社，1989 年），第 2 章；另请参阅史密森尼视频历史计划，兰德公司访谈，1990 年 6 月 12-13 日；与 Clifford Shaw 的访谈，1990 年 6 月 12 日，第 13 页。

20. Paul Ceruzzi, Beyond the Limits: Flight Enters the Computer Age (Cambridge: MIT Press, 1989), chapter 2; see also Smithsonian Videohistory Program, RAND Corporation interviews, June 12–13, 1990; interview with Clifford Shaw, 12 June 1990, 13.

21.在法国，Compagnie des Machines Bull 于 1952 年推出了一款具有类似架构的机器。这款机器被称为“Gamma 3”，非常成功，是法国生产的首批销量达到临界水平的产品之一。请参阅 Bruno LeClerc 的“从 Gamma 2 到 Gamma ET：Bull 电子计算的诞生”，《计算机历史年鉴》 12：1（1990 年）：5-22。

21. In France, Compagnie des Machines Bull introduced, in 1952, a machine having a similar architecture. Called the “Gamma 3,” it was very successful and was one of the first products produced in France to achieve a critical mass of sales. See Bruno LeClerc, “From Gamma 2 to Gamma E.T.: The Birth of Electronic Computing at Bull,” Annals of the History of Computing 12: 1 (1990): 5–22.

22.例如，请参阅计算机研究公司，《卡片编程计算机与通用模型 CRC 102A 的比较》，16 页小册子（1953 年），国家航空航天博物馆，NBS 档案。

22. See, for example, Computer Research Corporation, “Comparison of the Card-Programmed Computer [sic] with the General-Purpose Model CRC 102A,” 16 page pamphlet (1953) National Air & Space Museum, NBS archive.

23. David Alan Grier，《ENIAC、动词‘编程’和数字计算机的出现》，《计算机历史年鉴》 18：1（1996 年）：51-55。

23. David Alan Grier, “The ENIAC, the Verb ‘to program’ and the Emergence of Digital Computers,” Annals of the History of Computers 18: 1 (1996): 51–55.

24. “历史评论”，载于LR Johnson著《数据、程序和计算机的系统结构》（新泽西州恩格尔伍德克利夫斯：普伦蒂斯霍尔出版社，1970年）：185。原始备忘录的副本现存于宾夕法尼亚大学档案馆。

24. “Historical Comments,” in L. R. Johnson, System Structure in Data, Programs, and Computers (Englewood Cliffs, NJ: Prentice-Hall, 1970): 185. A copy of the original memorandum is in the University of Pennsylvania archives.

25.有关 EDVAC 命运的讨论，请参阅 Michael Williams，《EDVAC 的起源、用途和命运》，《计算史年鉴》第 15 卷（1993 年）：22-38。

25. For a discussion of the fate of the EDVAC see Michael Williams, “The Origins, Uses, and Fate of the EDVAC,” Annals of the History of Computing 15 (1993): 22–38.

26.第一稿的副本现存于美国国家航空航天博物馆档案馆，NBS 收藏。

26. A copy of the First Draft is in the National Air and Space Museum Archives, NBS Collection.

27.这也源于 IAS 机器在多个地方被复制的事实。

27. It also comes from the fact that the IAS machine was copied in a number of locations.

28. Herman Goldstine，《从帕斯卡到冯·诺依曼的计算机》（普林斯顿：普林斯顿大学出版社，1972）：182。

28. Herman Goldstine, The Computer from Pascal to von Neumann (Princeton: Princeton University Press, 1972): 182.

29.由于“编程”这一术语尚未开始使用，因此在本节中我使用“设置”。

29. Because the term “programming” had not yet come into use, I use “set up” in this section.

30. John Mauchly，《EDVAC 型机器的问题准备》，哈佛大学，《大型数字计算机器研讨会论文集》（剑桥：哈佛大学出版社，1948 年）：第 203-207 页。

30. John Mauchly, “Preparation of Problems for EDVAC-Type Machines,” Harvard University, Proceedings of a Symposium on Large-Scale Digital Calculating Machinery (Cambridge: Harvard University Press, 1948): 203–207.

31. Eckert，《Disclosure...》，写于 1944 年 1 月 29 日；重印于 Herman Lukoff 所著的《从点到位：电子计算机的个人历史》（波特兰，俄勒冈州：机器人出版社）：第 207-209 页。

31. Eckert, “Disclosure ... ,” written January 29, 1944; reprinted in Herman Lukoff, From Dits to Bits: A Personal History of the Electronic Computer (Portland, OR: Robotics Press): 207–209.

32. J. Presper Eckert，《磁力计算机的披露》，宾夕法尼亚大学摩尔电气工程学院，1944 年 1 月 29 日备忘录；载于 Lukoff 的《从点到位》，第 207-209 页；J. Presper Eckert 和 John Mauchly 的《自动高速计算：EDVAC 的进展报告》部分内容转载于 Johnson 的《数据、程序和计算机中的系统结构》，第 184-187 页。关于冯·诺依曼与 Eckert 和 Mauchly 之间的关系，以及他们各自对 EDVAC 报告的相对贡献，有许多记载。参见 Goldstine 的《从帕斯卡到冯·诺依曼的计算机》；Mauchly 本人的记载见于《修正 ENIAC 故事》，Datamation（1979 年 10 月）：第 217-220 页。这些事件的细节在 1974 年结束的霍尼韦尔诉斯佩里兰德公司 (Honeywell v. Sperry Rand, Inc.)案的审判中有所提及。

32. J. Presper Eckert, “Disclosure of a Magnetic Calculating Machine,” University of Pennsylvania, Moore School of Electrical Engineering, memorandum of January 29, 1944; in Lukoff, From Dits to Bits, 207–209; also J. Presper Eckert and John Mauchly, “Automatic High Speed Computing: A Progress Report on the EDVAC,” portions reprinted in Johnson, System Structure in Data, Programs, and Computers, 184–187. There are many accounts of the relationship of von Neumann with Eckert and Mauchly, and of the relative contributions each made to the EDVAC report. See Goldstine, Computer From Pascal to von Neumann; Mauchly’s own account is told in “Amending the ENIAC Story,” Datamation (October 1979): 217–220. The details of these events were covered in the trial Honeywell v. Sperry Rand, Inc., concluded in 1974.

33.上述讨论并未提及艾伦·图灵可能做出的贡献，他很久以前就以理论形式阐述了与此非常相似的原理。图灵可能启发了那些从事 EDVAC 设计的人，但目前尚不清楚。

33. The above discussion makes no mention of the possible contribution of Alan Turing, who stated something very much like this principle in theoretical terms long before. Turing may have inspired those working on the EDVAC design, but that is unknown at this time.

34.有关谁发明了存储程序原理的更多信息，请参阅《计算机史年鉴》第 4 卷（1982 年 10 月）：358-361 页。

34. For more on who invented the Stored-Program Principle, see Annals of the History of Computing 4 (October 1982): 358–361.

35.摩尔电气工程学院，《电子数字计算机设计理论与技术；1946 年 7 月 8 日至 8 月 31 日在摩尔电气工程学院举办的讲座》（剑桥：麻省理工学院出版社，1985 年[重印]）。

35. Moore School of Electrical Engineering, “Theory and Techniques for Design of Electronic Digital Computers; Lectures Given at the Moore School of Electrical Engineering, July 8–August 31, 1946” (Cambridge: MIT Press, 1985 [reprint]).

36. EDSAC 是“电子延迟存储自动计算器”的缩写；BINAC 是“二进制自动计算机”的缩写。

36. The acronym EDSAC stands for “Electronic Delay Storage Automatic Calculator”; BINAC stands for “Binary Automatic Computer.”

37.用于指计算机处理的一组数字的“字”这个术语可能也是来自冯·诺依曼。

37. The term “word” as applied to a chunk of digits handled in a computer probably also came from von Neumann.

38.该架构的典型现代描述在“数字计算机架构”中有所描述，载于 Jack Belzer、Albert Holzman 和 Allen Kent 编的《计算机科学与技术百科全书》第 7 卷（纽约：Dekker，1970 年）：289-326。

38. A typical modern description of the architecture is described in “Digital Computer Architecture,” in Jack Belzer, Albert Holzman, and Allen Kent, eds., Encyclopedia of Computer Science and Technology, vol. 7 (New York: Dekker, 1970): 289–326.

39.有关大学入门教科书如何处理这一主题的示例，请参阅 Helene G. Kershner 的《计算机素养导论》（马萨诸塞州莱克星顿：DC Heath，1990 年），Jack B. Rochester 和 Jon Rochester 的《为人民服务的计算机》（伊利诺伊州霍姆伍德：Richard D. Irwin，1991 年）。另请参阅 W. Danniel Hillis 的《连接机器》，《科学美国人》（1987 年 6 月）：108-115，其中明确阐述了并行设计的“非冯·诺依曼”性质。具有讽刺意味的是，Hillis 错误地将 ENIAC 识别为顺序机器；事实上，ENIAC 是一台并行处理机器。

39. For an example of how this subject is treated in introductory college-level textbooks, see Helene G. Kershner, Introduction to Computer Literacy (Lexington, MA: D. C. Heath, 1990), Jack B. Rochester and Jon Rochester, Computers for People (Homewood, IL: Richard D. Irwin, 1991). See also W. Danniel Hillis, “The Connection Machine,” Scientific American (June 1987): 108–115, for an explicit statement of the “non-von-Neumann” nature of parallel designs. Ironically, Hillis incorrectly identifies the ENIAC as a sequential machine; in fact, the ENIAC was a parallel processing machine.

40. Alan Perlis，“编程警句”，ACM Sigplan Notices（1981年10月）：7-13。

40. Alan Perlis, “Epigrams on Programming,” ACM Sigplan Notices (October 1981): 7–13.

41.在下面的讨论中，我主要依赖 Nancy Stern 所著《从 ENIAC 到 UNIVAC：对 Eckert-Mauchly 计算机的评价》（马萨诸塞州贝德福德：Digital Press，1981 年）中的论点和数据，尤其是第 5 章。

41. In the following discussion I rely heavily on the arguments and data in Nancy Stern, From ENIAC to UNIVAC: an Appraisal of the Eckert-Mauchly Computers (Bedford, MA: Digital Press, 1981), especially chapter 5.

42.莫奇利，1948 年 3 月 31 日备忘录；哈格利博物馆，斯佩里 Univac 公司记录，系列 I，盒 3，文件夹“埃克特-莫奇利计算机公司；莫奇利，约翰。”

42. Mauchly, memorandum of 3=31=1948; Hagley Museum, Sperry Univac Company Records, Series I, Box 3, folder “Eckert-Mauchly Computer Corporation; Mauchly, John.”

43.冯·诺依曼 1945 年 5 月 8 日写给赫尔曼·戈德斯坦的信；哈格利博物馆，斯佩里·尤尼瓦克公司记录；系列 II，盒子 74，文件夹“1945 年 5 月 - 1945 年 10 月”。

43. Von Neumann to Herman Goldstine, letter of May 8, 1945; Hagley Museum, Sperry Univac Corporate Records; Series II, Box 74, folder “May 1945–October 1945.”

44. Howard H. Aiken，《自动计算机的未来》，《电子研究与信息处理；NTZ 论文集》，第 4 组，1956 年，第 31-35 页。令人难以置信的是，他在 1956 年发表了这一声明，而那时已经有充分的证据表明，这种机器不仅存在，而且变得相当普遍。华莱士·埃克特在 20 世纪 40 年代末在哈佛举行的几次高速计算机会议上就计算机与穿孔卡机进行了评论。

44. Howard H. Aiken, “The Future of Automatic Computing Machinery,” Elektronische Rechenmaschinen und Informationsverarbeitung; Beihefte der NTZ, Band 4, 1956, pp. 31–35. Incredibly, he made this statement in 1956, by which time there was ample evidence that such machines not only existed but were becoming rather common. Wallace Eckert commented on computers vs. punched card machines at one of several conferences on high speed computing machinery held at Harvard in the late 1940s.

45. Stern，《从 ENIAC 到 UNIVAC》，第 91 页。Stern 认为“这两个人……事实上被解雇了。”

45. Stern, From ENIAC to UNIVAC, 91. Stern believes that “the two men ... were, in fact, fired.”

46.莫奇利致 JP Eckert Jr. 等人，1948 年 1 月 12 日；哈格利博物馆，Sperry UNIVAC 公司记录，系列 I，盒 3，文件夹“Eckert-Mauchly 计算机公司；莫奇利，约翰。”

46. Mauchly to J. P. Eckert Jr. et al, 1=12=1948; Hagley Museum, Sperry UNIVAC Company Records, Series I, Box 3, folder “Eckert-Mauchly Computer Corporation; Mauchly, John.”

47. 计算机及其未来(威尔士拉迪德诺，1970 年)：7-8

47. Computers and Their Future (Lladudno, Wales, 1970): 7–8

48. Stern，《从 ENIAC 到 UNIVAC》，第 148–151 页。在英国，LEO 计算机于 1951 年 2 月在伦敦 J. Lyons Catering Company 的场地运行测试程序，比 UNIVAC 交付早一个月。然而，直到 1953 年底，LEO 才被认为完成。参见 SH Lavington，《早期英国计算机》（马萨诸塞州贝德福德：Digital Press，1980 年）：第 72–73 页；另见 Peter Bird，《LEO，里昂的骄傲》，《计算机历史年鉴》第 14 卷：第 2 期（1992 年）：第 55–56 页。

48. Stern, From ENIAC to UNIVAC, 148–151. In Britain, the LEO computer ran test programs on the premises of the J. Lyons Catering Company of London by February 1951, a month before the UNIVAC delivery. However, it was not until late 1953 before LEO was considered finished. See S. H. Lavington, Early British Computers (Bedford, MA: Digital Press, 1980): 72–73; also Peter Bird, “LEO, the Pride of Lyons,” Annals of the History of Computing 14: 2 (1992): 55–56.

49.口述历史会议，UNISYS 公司，1990 年 5 月 17-18 日，史密森学会，华盛顿特区。

49. Oral History Session, UNISYS Corporation, May 17–18, 1990, Smithsonian Institution, Washington, DC.

50. LR Johnson，《大型电子计算机的安装》，ACM 会议论文集（多伦多，1952 年 9 月 8-10 日）：77-81。

50. L. R. Johnson, “Installation of a Large Electronic Computer,” in Proceedings of the ACM Meeting (Toronto, September 8–10, 1952): 77–81.

51. J. Presper Eckert，“关于计算历史的思考”，IEEE 计算机（1976 年 12 月）：64。

51. J. Presper Eckert, “Thoughts on the History of Computing,” IEEE Computer (December 1976): 64.

52. Luther A. Harr，《Univac 系统，1954 年进展报告》 (雷明顿·兰德公司，1954 年)：第 6 页。

52. Luther A. Harr, “The Univac System, a 1954 Progress Report” (Remington Rand Corporation, 1954): 6.

53. James C. McPherson，《操作 UNIVAC 系统的人口普查经验》，《数字计算机安装管理方面研讨会》（1953 年 3 月 30 日，华盛顿特区）：第 33 页。

53. James C. McPherson, “Census Experience Operating a UNIVAC System,” Symposium on Managerial Aspects of Digital Computer Installations (30 March, 1953, Washington, DC): 33.

54. Lukoff，《从 Dits 到 Bits》，第 9 章。

54. Lukoff, From Dits to Bits, chapter 9.

55. Roddy F. Osborn，《GE 和 UNIVAC：驾驭高速计算机》，《哈佛商业评论》（1954 年 7-8 月）：102。

55. Roddy F. Osborn, “GE and UNIVAC: Harnessing the High-Speed Computer,” Harvard Business Review (July–August 1954): 102.

56. McPherson，“操作 UNIVAC 系统的人口普查经验”，第 30-36 页。

56. McPherson, “Census Experience Operating a UNIVAC System,” 30–36.

57.雷明顿·兰德公司，“Univac Fac-Tronic 系统”，未注明日期的手册，约 1951 年，Unisys 档案馆；另请参阅 McPherson，“操作 UNIVAC 系统的人口普查经验”。

57. Remington Rand Corporation, “Univac Fac-Tronic System,” Undated brochure, ca. 1951, Unisys Archives; also McPherson, “Census Experience Operating a UNIVAC System.”

58. Paul E. Ceruzzi，“第一代计算机和航空航天工业”，美国国家航空航天博物馆研究报告（1985 年）：75-89；另请参阅 Robert Dorfman，“线性规划的发现”，计算机史年鉴6：3（1984 年）：283-295，以及 Mina Rees，“海军研究办公室的计算计划”，计算机史年鉴4：2（1982 年）：102-120。

58. Paul E. Ceruzzi,“The First Generation of Computers and the Aerospace Industry,” National Air and Space Museum Research Report (1985): 75–89; also Robert Dorfman, “The Discovery of Linear Programming,” Annals of the History of Computing 6: 3 (1984): 283–295, and Mina Rees, “The Computing Program at the Office of Naval Research,” Annals of the History of Computing 4: 2 (1982): 102– 120.

59. LR Johnson，“大型电子计算机的安装”，ACM 会议论文集（多伦多，1952 年）：77-81。

59. L. R. Johnson, “Installation of a Large Electronic Computer,” In Proceedings of ACM Meeting (Toronto, 1952): 77–81.

60. Luther Harr，《UNIVAC 系统，1954 年进展报告》 (雷明顿·兰德公司，1954 年)，UNISYS 档案。

60. Luther Harr, “The UNIVAC System, a 1954 Progress Report,” (Remington Rand Corporation, 1954) UNISYS Archives.

61.同上，第7页。

61. Ibid., 7.

62.劳伦斯利弗莫尔实验室，《劳伦斯利弗莫尔实验室的计算》，UCID 报告 20079，1984 年。

62. Lawrence Livermore Laboratory, “Computing at Lawrence Livermore Laboratory,” UCID Report 20079, 1984.

63. Roddy F. Osborn，《GE 和 UNIVAC：驾驭高速计算机》，《哈佛商业评论》（1954 年 7-8 月）：99-107。

63. Roddy F. Osborn, “GE and UNIVAC: Harnessing the High-Speed Computer,” Harvard Business Review (July–August 1954): 99–107.

64. John Diebold，《自动化》（纽约：Van Nostrand，1952 年）。

64. John Diebold, Automation (New York: Van Nostrand, 1952).

65. John Diebold，《无人工厂：新工业革命》，《国家》（1953 年）：第 227–228 页、第 250–251 页、第 271–272 页。另请参阅 David F. Noble，《生产力：工业自动化的社会史》（纽约：Knopf，1986 年），第 4 章。

65. John Diebold, “Factories Without Men: New Industrial Revolution,” The Nation (1953): 227–228, 250–251, 271–272. See also David F. Noble, Forces of Production: A Social History of Industrial Automation (New York: Knopf, 1986), chapter 4.

66. Roddy Osborn，《GE与UNIVAC》，第99页。

66. Roddy Osborn, “GE and UNIVAC,” 99.

67.同上，第 103 页；另见史密森学会档案馆的《UNIVAC 口述历史》。

67. Ibid., 103; also UNIVAC Oral History, Smithsonian Institution Archives.

68.同上，第104页。

68. Ibid., 104.

69.同上，第107页。

69. Ibid., 107.

70. Harr，“UNIVAC 系统：1954 年进展报告”，1。UNISYS 档案。

70. Harr, “The UNIVAC System: A 1954 Progress Report,” 1. UNISYS Archives.

71.史密森尼=UNISYS UNIVAC 口述历史项目，1990 年 5 月 17-18 日。

71. Smithsonian=UNISYS UNIVAC Oral History Project, May 17–18 (1990).

72. Emerson W. Pugh，《塑造一个行业的记忆》（剑桥：麻省理工学院出版社，1984 年）：第 30 页。

72. Emerson W. Pugh, Memories that Shaped an Industry (Cambridge: MIT Press, 1984): 30.

73. Saul Rosen，《电子计算机：历史综述》，《计算综述》 1（1969 年 3 月）：7-36。

73. Saul Rosen, “Electronic Computers: a Historical Survey,” Computing Surveys 1 (March 1969): 7–36.

74. Bashe 等人，IBM 的早期计算机，161-162。

74. Bashe et al., IBM’s Early Computers, 161–162.

75. Cuthbert C. Hurd，《编辑的证词》，《计算机史年鉴》第 3 卷（1981 年 4 月）：168。

75. Cuthbert C. Hurd, “Edited Testimony,” Annals of the History of Computing 3 (April 1981): 168.

76.同上，第169页。

76. Ibid., 169.

77.同上，第169页。

77. Ibid., 169.

78.同上，第169页。

78. Ibid., 169.

79. Bashe 等人，IBM 的早期计算机129。

79. Bashe et al., IBM’s Early Computers 129.

80.同上，第 173–178 页。

80. Ibid., 173–178.

81. Erwin Tomash 和 Arnold A. Cohen。“ERA 的诞生：工程研究协会，1946-1955”，《计算史年鉴》 1：2（1979 年）：83-97；另请参阅 Charles J. Murray 的《超人：Seymour Cray 和超级计算机背后的技术奇才的故事》（纽约：Wiley，1997 年）。

81. Erwin Tomash and Arnold A. Cohen. “The Birth of an ERA: Engineering Research Associates, Inc., 1946–1955,” Annals of the History of Computing 1: 2 (1979): 83–97; also Charles J. Murray, The Supermen: The Story of Seymour Cray and the Technical Wizards behind the Supercomputer (New York: Wiley, 1997).

82. Seymour Cray，《砷化镓到底是怎么回事？》，作者所拥有的录像带，记录了他在 1988 年 11 月于佛罗里达州奥兰多举行的“超级计算 '88”会议上的演讲；另请参阅 Murray 的《超人》，第 44–45 页。

82. Seymour Cray, “What’s All This About Gallium Arsenide?” Videotape in the author’s possession of a talk given at the “Supercomputing ’88” conference in Orlando, Florida, November 1988; also Murray, The Supermen, 44–45.

83. Tomash 和 Cohen，《ERA 的诞生》，第 90 页。

83. Tomash and Cohen, “The Birth of an ERA,” 90.

84. Seymour R. Cray，“ERA 1103 计算机系统内部存储器部分的计算机编程预防性维护”，WESCON 论文集（1954 年）：62-66 页。

84. Seymour R. Cray, “Computer-Programmed Preventative Maintenance for Internal Memory Sections of the ERA 1103 Computer System,” in Proceedings of WESCON (1954): 62–66.

85. Samuel S. Snyder，“美国密码组织对数字计算机产业的影响”，系统与软件杂志1 (1979): 87-102。

85. Samuel S. Snyder, “Influence of U.S. Cryptologic Organizations on the Digital Computer Industry,” Journal of Systems and Software 1 (1979): 87–102.

86. Ben Ferber，《Charactron 与 ERA 1103 的使用》，《WJCC 论文集》（1956 年 2 月 7-9 日）：第 34-35 页。

86. Ben Ferber, “The Use of the Charactron with the ERA 1103,” Proceedings of WJCC (February 7–9, 1956): 34–35.

87. Alice R. 和 Arthur W. Burks，《第一台电子计算机：阿塔纳索夫的故事》（安阿伯：密歇根大学出版社，1988 年），第 1 章；另请参阅 J. Presper Eckert 的《数字计算机存储系统概述》，载于《IRE 论文集》（1953 年 10 月）：1393-1406 页。

87. Alice R. and Arthur W. Burks, The First Electronic Computer: The Atanasoff Story (Ann Arbor: University of Michigan Press, 1988), chapter 1; also J. Presper Eckert, “A Survey of Digital Computer Memory Systems,” in Proceedings of IRE (October 1953): 1393–1406.

88. Perry O. Crawford Jr.，《通过算术运算实现自动控制》（硕士论文，麻省理工学院，1942 年）。Crawford 的论文主要讨论了照相存储技术而非磁存储技术，但这篇论文似乎在一定程度上启发了 Eckert 后来在 Moore 学院所做的工作。

88. Perry O. Crawford Jr., “Automatic Control by Arithmetic Operations” (Master’s thesis, MIT, 1942). The bulk of Crawford’s thesis discussed photographic rather than magnetic storage techniques, but it appears that this thesis was partially the inspiration for work that Eckert later did at the Moore School.

89. Tomash 和 Cohen，《ERA 的诞生》，第 83-97 页。

89. Tomash and Cohen, “The Birth of an ERA,” 83–97.

90.工程研究协会，高速计算设备(纽约：McGraw Hill，1950 年；1983 年重印，洛杉矶：Tomash Publishers)：322–339。

90. Engineering Research Associates, Inc., High-Speed Computing Devices (New York: McGraw Hill, 1950; reprinted 1983, Los Angeles: Tomash Publishers): 322– 339.

91. Richard E. Sprague，《CADAC》，美国海军，中等价位商用通用电子数字计算机研讨会（华盛顿特区：1952 年）：13-17。

91. Richard E. Sprague, “The CADAC,” U.S. Navy, Symposium on Commercially Available General-Purpose Electronic Digital Computers of Moderate Price (Washington, DC: 1952): 13–17.

92. Richard E. Sprague，“西方视角下的计算机历史”，Comm ACM 15（1972 年 7 月）：686-692。

92. Richard E. Sprague, “A Western View of Computer History,” Comm ACM 15 (July 1972): 686–692.

93. ED Lucas，“图形数据与数字计算机的有效链接”，WESCON 论文集（1954 年）：32-37。

93. E. D. Lucas, “Efficient Linkage of Graphical Data with a Digital Computer,” in Proceedings of WESCON (1954): 32–37.

94. Willis E. Dobbins，《设计低成本通用计算机》，ACM 会议论文集（多伦多，1952 年 9 月）：第 28-29 页。

94. Willis E. Dobbins, “Designing a Low-Cost General Purpose Computer,” in Proceedings of ACM Meeting (Toronto, September, 1952): 28–29.

95. Sprague，“西方视角下的计算机历史”，CACM：686-692。

95. Sprague, “A Western View of Computer History,” CACM: 686–692.

96. C. Gordon Bell 和 Allen Newell，《计算机结构：阅读和实例》（纽约：McGraw-Hill，1971）：217。

96. C. Gordon Bell and Allen Newell, Computer Structures: Readings and Examples (New York: McGraw-Hill, 1971): 217.

97. BE Carpenter 和 RW Doran，《另一台图灵机》，《计算机期刊》20（1977 年 8 月）：269-279。

97. B. E. Carpenter and R. W. Doran, “The Other Turing Machine,” Computer Journal 20 (August 1977): 269–279.

98. Martin Campbell-Kelly，《对试点 ACE 进行编程：国家物理实验室的早期编程》，《计算史年鉴》第 3 卷（1981 年）：133-162。

98. Martin Campbell-Kelly, “Programming the Pilot ACE: Early Programming at the National Physical Laboratory,” Annals of the History of Computing 3 (1981): 133–162.

99. Bashe 等人，《IBM 的早期计算机》，168。

99. Bashe et al., IBM’s Early Computers, 168.

100. Tomash 和 Cohen，《ERA 的诞生》，第 83–97 页。

100. Tomash and Cohen, “The Birth of an ERA,” 83–97.

101. Bashe 等人，《IBM 的早期计算机》，第 3 章和第5章。

101. Bashe et al., IBM’s Early Computers, chapters 3 and 5.

102.同上，第170–172页。

102. Ibid., 170–172.

103. Thomas J. Watson Jr.，《父、子与公司：我在 IBM 的一生及以后》（纽约：Bantam Books，1990 年）：第 224 页。

103. Thomas J. Watson Jr., Father, Son & Co: My Life at IBM and Beyond (New York: Bantam Books, 1990): 224.

104. Lancelot W. Armstrong，UNIVAC 会议，史密森学会，1990 年 5 月 17-18 日（华盛顿特区：史密森学会档案馆）：105。

104. Lancelot W. Armstrong, UNIVAC Conference, Smithsonian Institution, May 17–18, 1990 (Washington, DC: Smithsonian Institution Archives): 105.

105. FC Williams 和 T. Kilburn，“用于二进制数字计算机的存储系统”，电气工程师学会会刊 III：96（1949 年 3 月）：81-100。

105. F. C. Williams and T. Kilburn, “A Storage System for Use with Binary-Digital Computing Machines,” Institution of Electrical Engineers in Proceedings III: 96 (March 1949): 81–100.

106. JC Chu 和 RJ Klein，“威廉姆斯电子管选择程序”，ACM 会议论文集（多伦多，1952 年 9 月）：110-114。

106. J. C. Chu and R. J. Klein, “Williams Tube Selection Program,” in Proceedings ACM Meeting (Toronto, September, 1952): 110–114.

第2章

Chapter 2

1. Joanne Yates，《通过沟通进行控制：美国管理体系的兴起》（巴尔的摩：约翰霍普金斯大学出版社，1989 年）。

1. Joanne Yates, Control Through Communication: The Rise of System in American Management (Baltimore: Johns Hopkins University Press, 1989).

2. Francis J. Murray，《数学机器》，第 1 卷：数字计算机（纽约：哥伦比亚大学出版社，1961 年）：43-44；另请参阅 Edwin Darby 的《一切都加起来：Victor Comptometer 公司的成长》（芝加哥：Victor Comptometer 公司，1968 年）。

2. Francis J. Murray, Mathematical Machines, vol. 1: Digital Computers (New York: Columbia University Press, 1961): 43–44; also Edwin Darby, It All Adds Up: The Growth of the Victor Comptometer Corporation (Chicago: Victor Comptometer Corporation, 1968).

3.这种特性被称为“磁滞”，源于希腊语，表示“不足”或“滞后”，因为材料磁化的速度落后于磁化电流影响它的速率。

3. The property is called “hysteresis,” after the Greek word to denotestext1 “deficiency” or “lagging,” because the rate at which the material becomes magnetized lags behind the rate at which a magnetizing current can affect it.

4. Emerson Pugh，《塑造一个行业的记忆：导致 IBM 系统 360 的决策》（剑桥：麻省理工学院出版社，1984 年），第 2 章。

4. Emerson Pugh, Memories that Shaped an Industry: Decisions Leading to IBM System=360 (Cambridge: MIT Press, 1984), chapter 2.

5. Kent C. Redmond 和 Thomas M. Smith，《旋风计划：先锋计算机的历史》（马萨诸塞州贝德福德：Digital Press，1980 年）：第 206 页；《ENIAC 静态磁存储器工程报告》，Burroughs Corporation，为费城阿伯丁试验场弹道研究实验室准备的报告，1953 年 7 月 31 日；另见《电子学》（1953 年 5 月）：第 198 页和第 200 页。史密森尼学会的美国国家历史博物馆收藏着使用王氏发明的 Mark IV 移位寄存器的残存件之一。

5. Kent C. Redmond and Thomas M. Smith, Project Whirlwind: The History of a Pioneer Computer (Bedford, MA: Digital Press, 1980): 206; “Engineering Report on the Static Magnetic Memory for the ENIAC,” Burroughs Corporation, Report prepared for the Ballistics Research Laboratory, Aberdeen Proving Ground, Philadelphia, 31 July, 1953; also Electronics (May 1953): 198 and 200. The Smithsonian’s National Museum of American History has one of the surviving remnants of a Mark IV shift register that used Wang’s invention.

6.Pugh，《回忆录》，59页。

6. Pugh, Memories, 59.

7. “SAGE（半自动地面环境）”，特刊，计算机史年鉴5：4（1983 年）。

7. “SAGE (Semi-Automatic Ground Environment),” Special Issue, Annals of the History of Computing 5: 4 (1983).

8. Morton M. Astrahan 和 John F. Jacobs，《SAGE 计算机设计史——AN=FSQ-7》，《计算机历史年鉴》 5：4（1983）：343–344。一位目击者回忆道，当团队参观 IBM 工厂时，一位正在调整磁鼓的技术人员不小心让螺丝刀碰到了磁鼓表面，发出的声音就像指甲划过黑板的声音。令所有人惊讶的是，磁鼓上存储的数据没有丢失。无论如何，IBM 为其商用计算机建立的有序生产设施被认为领先于其他公司。

8. Morton M. Astrahan and John F. Jacobs, “History of the Design of the SAGE Computer–-the AN=FSQ-7,” Annals of the History of Computing 5: 4 (1983): 343– 344. One witness to those events remembered that as the team was visiting an IBM facility, a technician who was adjusting a drum accidentally let the screwdriver hit the drum’s surface, causing a sound not unlike fingernails run across a blackboard. To everyone’s surprise, no data stored on the drum were lost. In any event, the orderly production facilities that IBM had set up for its commercial computers was judged to be ahead of what the other companies could show.

9. Gordon Bell，“计算机博物馆会员的首次实地考察：Northbay [原文如此] AN=FSQ SAGE 站点”，CACM 26:2（1983）：118-119。

9. Gordon Bell, “The Computer Museum Member’s First Field Trip: The Northbay [sic] AN=FSQ SAGE Site,” CACM 26: 2 (1983): 118–119.

10. Edmund Van Deusen，《电子走向现代化》，《财富》（1955 年 6 月）：132–136、148。

10. Edmund Van Deusen, “Electronics Goes Modern,” Fortune (June 1955): 132– 136, 148.

11. Pugh，《回忆录》，102–117。

11. Pugh, Memories, 102–117.

12.同上，第126页。

12. Ibid., 126.

13. Katherine Fishman，《计算机机构》（纽约：Harper & Row，1981）：44。

13. Katherine Fishman, The Computer Establishment (New York: Harper & Row, 1981): 44.

14. Saul Rosen，《电子计算机：历史综述》，《计算综述》 1（1969 年 3 月）：7-36。

14. Saul Rosen, “Electronic Computers: a Historical Survey,” Computing Surveys 1 (March 1969): 7–36.

15.同上；另请参阅 Fishman 的《计算机机构》，第 161-162 页。有关 RAYDAC 的详细信息，请参阅 Engineering Research Associates 的《高速计算设备》（纽约，McGraw-Hill，1950 年）：第 206-207 页。

15. Ibid.; also Fishman, The Computer Establishment, 161–162. For details on the RAYDAC, see Engineering Research Associates, High Speed Computing Devices (New York, McGraw-Hill, 1950): 206–207.

16.要了解霍尼韦尔决策的内部人士观点，请参阅史密森尼计算机历史项目对理查德·布洛赫的采访；另请参阅 W. David Gardner 的《Chip off the Old Bloch》，Datamation（1982 年 6 月）：241-242。

16. For an insider’s view of Honeywell’s decision, see interview with Richard Bloch, Smithsonian Computer History Project; also W. David Gardner, “Chip off the Old Bloch,” Datamation (June 1982): 241–242.

17. 《财富》 52（1955 年 7 月），增刊，2-5。

17. Fortune 52 (July 1955), supplement, 2–5.

18. Robert W. House，“OARAC 上的可靠性体验”，载于《东部联合计算机会议论文集》（1953 年）：43-44；另请参阅 Robert Johnson 的访谈，《计算机史年鉴》 12：2（1990 年）：130-137；以及 George Snively，“通用电气进入计算机行业”，《计算机史年鉴》 10：（1988 年）：74-78。

18. Robert W. House, “Reliability Experience on the OARAC,” in Proceedings Eastern Joint Computer Conference (1953): 43–44; also Robert Johnson, Interview, Annals of the History of Computing 12: 2 (1990): 130–137; and George Snively, “General Electric Enters the Computer Business, Annals of the History of Computing 10: (1988): 74–78.

19. Homer R. Oldfield，《七个小矮人之王：通用电气对计算机行业的暧昧挑战》（加利福尼亚州洛斯阿拉米托斯：IEEE 计算机协会出版社，1996 年）。

19. Homer R. Oldfield, King of the Seven Dwarfs: General Electric’s Ambiguous Challenge to the Computer Industry (Los Alamitos, CA: IEEE Computer Society Press, 1996).

20. Fishman，《计算机机构》，164-165。

20. Fishman, Computer Establishment, 164–165.

21. WK Halstead 等人，《RCA BIZMAC 系统的目的和应用》，《西部联合计算机会议论文集》（1956 年）：119-123；另见 Rosen，《电子计算机》，第 16-17 页。

21. W. K. Halstead et al., “Purpose and Application of the RCA BIZMAC System,” in Proceedings Western Joint Computer Conference (1956): 119–123; also Rosen, “Electronic Computers,” 16–17.

22. Rosen，“电子计算机”，16-17；WK Halstead 等人，“RCA BIZMAC 系统的目的和应用”，119-123。

22. Rosen, “Electronic Computers,” 16–17; W. K. Halstead, et al., “Purpose and Application of the RCA BIZMAC System,” 119–123.

23. RP Daly，“使用 UNIVAC 文件计算机进行集成数据处理”，西部联合计算机会议论文集(1956 年)：95-98。

23. R. P. Daly, “Integrated Data Processing with the UNIVAC File Computer,” Proceedings Western Joint Computer Conference (1956): 95–98.

24. Franklin Fisher、James McKie 和 Richard Mancke，《IBM 和美国数据处理行业》 (纽约，Praeger，1983 年）：53。

24. Franklin Fisher, James McKie, and Richard Mancke, IBM and the U.S. Data Processing Industry (New York, Praeger, 1983): 53.

25.例如，请参阅 TA Heppenheimer，《冯·诺依曼如何指引道路》，《美国发明与技术遗产》（1990 年秋季）：8-16。

25. See, for example, T. A. Heppenheimer, “How von Neumann Showed the Way,” American Heritage of Invention & Technology (Fall 1990): 8–16.

26.下面的描述取自几篇教材，包括 John L. Hennessy 和 David A. Patterson 的《计算机体系结构：定量方法》（加利福尼亚州圣马特奥：Morgan Kaufmann，1990 年）；Simon Lavington 的《早期英国计算机》（马萨诸塞州贝德福德：Digital Press，1980 年）：106-119；C. Gordon Bell 和 Allen Newell 的《计算机结构：阅读和范例》（纽约：McGraw-Hill，1971 年）。

26. The following description is taken from several texts, including John L. Hennessy and David A. Patterson, Computer Architecture: a Quantitative Approach (San Mateo, CA: Morgan Kaufmann, 1990); Simon Lavington, Early British Computers (Bedford, MA: Digital Press, 1980): 106–119; C. Gordon Bell and Allen Newell, Computer Structures: Readings and Examples (New York: McGraw-Hill, 1971).

27. Bell 和 Newell，《计算机结构》，224；此外，Adams Associates 的“计算机特性图表” Datamation（1960 年 11 月 =12 月）：14-17。

27. Bell and Newell, Computer Structures, 224; also Adams Associates, “Computer Characteristics Chart,” Datamation (November=December, 1960): 14–17.

28. Simon Lavington，《曼彻斯特计算机史》（英国曼彻斯特：NCC 出版物，1975 年）：12。

28. Simon Lavington, History of Manchester Computers (Manchester, UK: NCC Publications, 1975): 12.

29. CC Hurd，《早期 IBM 计算机：编辑过的证词》，计算机史年鉴3：2（1981 年 4 月）：185-176。

29. C. C. Hurd, “Early IBM Computers: Edited Testimony,” Annals of the History of Computing 3: 2 (April 1981): 185–176.

30. Lavington，《曼彻斯特计算机史》，78-82；Hennessy 和 Patterson，《计算机体系结构》，91-92。

30. Lavington, History of Manchester Computers, 78–82; also Hennessy and Patterson, Computer Architecture, 91–92.

31. Werner Buchholz 编，规划计算机系统：延伸项目（纽约：McGraw-Hill，1962 年），第 9 章。

31. Werner Buchholz, ed., Planning a Computer System: Project Stretch (New York: McGraw-Hill, 1962), chapter 9.

32.计算机体系结构中的堆栈概念可能起源于慕尼黑工业大学的 FL Bauer 教授；例如，参见 FL Bauer 的“公式控制的逻辑计算机 STANISLAUS”，MTAC 14（1960）：64-67。

32. The stack concept in computer architecture probably originated with Professor F. L. Bauer of the Technical University of Munich; see for example, F. L. Bauer, “The Formula-Controlled Logical Computer STANISLAUS,” MTAC 14 (1960): 64–67.

33. Richard E. Smith，“计算机体系结构的历史概述”，计算机史年鉴10：4（1989）：286。

33. Richard E. Smith, “A Historical Overview of Computer Architecture,” Annals of the History of Computing 10: 4 (1989): 286.

34. TH Myer 和 IE Sutherland，《论显示处理器的设计》，《ACM 通讯》 11：6（1968 年 6 月）：410-414；另请参阅 C. Gordon Bell、J. Craig Mudge 和 John McNamara，《计算机工程：DEC 对硬件系统设计的视角》（马萨诸塞州贝德福德：Digital Press，1978）：202。

34. T. H. Myer and I. E. Sutherland, “On the Design of Display Processors,” Communications of the ACM 11: 6 (June 1968): 410–414; also C. Gordon Bell, J. Craig Mudge, and John McNamara, Computer Engineering: a DEC View of Hardware Systems Design (Bedford, MA: Digital Press, 1978): 202.

35. Bell，Mudge 和 McNamara，《计算机工程》，256-257。

35. Bell, Mudge, and McNamara, Computer Engineering, 256–257.

36. Gerald Brock，《电信业：市场结构的动态》（马萨诸塞州剑桥：Ballinger，1981 年）：187-194。诉讼很少与晶体管有关；但很大程度上涉及贝尔的西部电气子公司作为桑迪亚公司的管理者的角色，桑迪亚公司是位于新墨西哥州的军事基地，制造原子弹。

36. Gerald Brock, The Telecommunications Industry: the Dynamics of Market Structure (Cambridge, MA: Ballinger, 1981): 187–194. Little of the litigation had to do with the transistor; much, however, concerned Bell’s Western Electric subsidiary’s role as manager of the Sandia Corporation, a military installation in New Mexico that manufactured atomic bombs.

37. Thomas J. Misa，《军事需求、商业现实和晶体管的发展，1948-1958》，收录于 Merritt Roe Smith 主编的《军事企业和技术变革》（剑桥：麻省理工学院出版社，1985 年）：第 253-287 页；另收录于 JH Felker，《TRADIC 系统的性能》，收录于《东部联合计算机会议论文集》（1954 年）：第 46-49 页。这些机器使用的是点接触晶体管，这种晶体管不仅不可靠，而且难以大批量生产。

37. Thomas J. Misa, “Military Needs, Commercial Realities, and the Development of the Transistor, 1948–1958,” in Merritt Roe Smith, ed., Military Enterprise and Technological Change (Cambridge: MIT Press, 1985): 253–287; also J. H. Felker, “Performance of TRADIC System,” in Proceedings Eastern Joint Computer Conference (1954): 46–49. These machines were used point-contact transistors, which were not only unreliable but difficult to produce in large quantities.

38. John Allen，《TX-0：过去和现在》，计算机博物馆报告第 8 期（1984 年春季）：2-11。

38. John Allen, “The TX-0: its Past and Present,” Computer Museum Report #8 (Spring 1984): 2–11.

39. Samuel Snyder，“美国密码组织对数字计算机产业的影响”，系统与软件杂志1 (1979): 87-102。

39. Samuel Snyder, “Influence of U.S. Cryptologic Organizations on the Digital Computer Industry,” Journal of Systems and Software 1 (1979): 87–102.

40.同上；还有 1995 年 3 月 31 日与 Ray Potts 的电话交谈。

40. Ibid.; also telephone conversation with Ray Potts, March 31, 1995.

41. Herman Lukoff，《从点到位：电子计算机的个人历史》（波特兰，俄勒冈州：机器人出版社，1979 年）；JL Maddox 等，“TRANSAC S-1000 计算机”，《东部联合计算机会议论文集》（1956 年 12 月）：13-16；另请参阅 Saul Rosen，“电子计算机：历史调查”。

41. Herman Lukoff, From Dits to Bits: a Personal History of the Electronic Computer (Portland, OR: Robotics Press, 1979); J. L. Maddox et al., “The TRANSAC S-1000 Computer,” in Proceedings Eastern Joint Computer Conference (December 1956): 13– 16; also Saul Rosen, “Electronic Computers: a Historical Survey.”

42.同上，第89页。

42. Ibid., 89.

43. Fisher，《IBM与美国数据处理行业》，87。

43. Fisher, IBM and the U.S. Data Processing Industry, 87.

44. LP Robinson，《30-201 型电子数字计算机》，《中等价位商用通用数字计算机研讨会》（华盛顿特区，1952 年 5 月 14 日）：第 31-36 页。

44. L. P. Robinson, “Model 30-201 Electronic Digital Computer,“ Symposium on Commercially-available General-purpose Digital Computers of Moderate Price (Washington, DC, 14 May 1952): 31–36.

45. Fisher，《IBM与美国数据处理行业》，83页。

45. Fisher, IBM and the U.S. Data Processing Industry, 83.

46.多年来，这台计算机一直在史密森尼学会的美国国家历史博物馆展出。这一说法来自 Burroughs 为该展览提供的宣传册。史密森尼计算机历史项目存档了 Burroughs 几位员工的口述历史，其中包括 1972 年 4 月 11 日对罗伯特·坎贝尔的采访。

46. For many years this computer was on display at the Smithsonian Institution’s National Museum of American History. The claim is taken from a brochure Burroughs made available for that exhibit. Oral histories of several Burroughs employees are on file with the Smithsonian Computer History Project, including an interview with Robert Campbell, 11 April 1972.

47. William Rodgers，《思考：沃森一家和 IBM 的传记》（纽约，Stein & Day，1969 年）：58-60。

47. William Rodgers, Think: a Biography of the Watsons and IBM (New York, Stein & Day, 1969): 58–60.

48.同上，第213页。

48. Ibid., 213.

49.对 IBM 最激烈、最生动的攻击出现在 Ted Nelson 的《Computer Lib》（印第安纳州南本德：Ted Nelson，1974 年）：52-56 页。

49. The most colorful and vehement attack on IBM is found in Ted Nelson, Computer Lib (South Bend, IN: Ted Nelson, 1974): 52–56.

50. Eric Weiss，《讣告：Robert B. Forest》，《计算机史年鉴》 19：2（1997）：70-73。

50. Eric Weiss, “Obituary: Robert B. Forest,” Annals of the History of Computing 19: 2 (1997): 70–73.

51. Bashe 等人，《IBM 的早期计算机》，280。

51. Bashe et al., IBM’s Early Computers, 280.

52.同上，第286页。

52. Ibid., 286.

53.同上，另见 T. Noyes 和 WE Dickenson 的《磁盘随机存取存储器的工程设计》，西部联合计算机会议论文集（1956 年 2 月 7 日）：第 42-44 页。

53. Ibid., also T. Noyes and W. E. Dickenson, “Engineering Design of a Magnetic Disk Random Access Memory,” in Proceedings, Western Joint Computer Conference (February 7, 1956): 42–44.

54. Mitchell E. Morris，《RAMAC 教授的任期》，Datamation（1981 年 4 月）：195-198。

54. Mitchell E. Morris, “Professor RAMAC’s Tenure,” Datamation (April 1981): 195–198.

55.这个故事曾被几个人独立地告诉过作者，其中至少有一个人当时在 BMEWS 系统上工作。Charles Bashe 等人在 IBM 的官方历史中提到了“非常紧张的时间表”，但没有重复这个故事。请参阅 Bashe 等人的《IBM 的早期计算机》（剑桥：麻省理工学院出版社，1986 年）：第 446-449 页。

55. This story has been independently told to the author by several people, including at least one person who worked on the BMEWS system at the time. Charles Bashe, et al, in IBM’s official history, mentions a “very taut schedule” but does not repeat this story. See Bashe et al., IBM’s Early Computers (Cambridge: MIT Press, 1986): 446–449.

56.在其联邦系统部门，IBM 在其程序员工作的地点有两年的时间不允许使用交互式终端。所有程序都必须以标准表格形式提交。更典型的情况是让程序员使用表格，让打字员准备卡片组，但只提供一次 029 打孔，以便他或她可以不时打一两张卡片。参见 Robert N. Britcher，《卡片、信使和争取正确性》，《系统与软件杂志》第 17 期（1992 年）：281-284。

56. For a period of two years at its Federal Systems Division, IBM did not allow interactive terminals at the sites where its programmers were working. All programs had to be submitted on the standard forms. A more typical situation was to have the programmers use the forms, letting keypunchers prepare the decks, but providing a single 029 punch so that he or she could punch one or two cards from time to time. See Robert N. Britcher, “Cards, Couriers, and the Race to Correctness,” Journal of Systems and Software 17 (1992): 281–284.

57. Bashe 等人，IBM 的早期计算机，468–469。

57. Bashe et al., IBM’s Early Computers, 468–469.

58. Bell 和 Newell，《计算机结构》，第 18 章；Fisher 等人，《IBM 和美国数据处理行业》，第 53 页。

58. Bell and Newell, Computer Structures, chapter 18; also Fisher et al., IBM and the U.S. Data Processing Industry, 53.

第3章

Chapter 3

1.我感谢弗吉尼亚理工大学的 JAN Lee 教授提醒我注意这一引文。

1. I wish to thank Professor J. A. N. Lee, of Virginia Tech, for bringing this citation to my attention.

2.这个故事是兰德公司研究数学家梅里尔·弗勒德在给《数据科学》（1984 年 12 月 1 日）第 15-16 页的一封信中讲述的。弗勒德说，他是从别人那里听到艾克的评论的。

2. The story is told by Merrill Flood, a research mathematician at the RAND Corporation, in a letter to the editor, Datamation (December 1, 1984): 15–16. Flood says that he heard Ike’s comment secondhand.

3.美国国家工程院，华盛顿特区。1993 年 10 月 6 日新闻稿。

3. National Academy of Engineering, Washington, DC. Press release dated 6 October 1993.

4. Susan Lammers 编，《程序员在工作》（华盛顿州雷德蒙德：微软出版社，1989 年），第 278 页。

4. Susan Lammers, ed., Programmers at Work (Redmond, WA: Microsoft Press, 1989), 278.

5.玛格丽特·汉密尔顿，作者访谈，1992 年 4 月 6 日；J.大卫·博尔特，《图灵的人：计算机时代的西方文化》（教堂山：北卡罗来纳大学出版社，1984 年）。

5. Margaret Hamilton, interview with the author, 6 April 1992; J. David Bolter, Turing’s Man: Western Culture in the Computer Age (Chapel Hill: University of North Carolina Press, 1984).

6. Alan Perlis 曾说过：“小心图灵的焦油坑，一切皆有可能，但没有什么是容易的。”摘自 Perlis 的《编程警句》，ACM SIGPLAN 通告（1981 年 10 月）：第 10 页。

6. Alan Perlis has said of this, “Beware of the Turing tar-pit in which everything is possible but nothing is easy.” In Perlis, “Epigrams on Programming,” ACM SIGPLAN Notices (October 1981): 10.

7. Barry Boehm，《软件工程》，IEEE 计算机学报C25（1976 年 12 月），1226-1241 页；有关 Boehm 论文的驳斥，请参阅 Werner Frank，《神话的历史 #1》，Datamation（1983 年 5 月）：252-263 页。

7. Barry Boehm, “Software Engineering,” IEEE Transactions on Computers C25 (December 1976), 1226–1241; for a refutation of Boehm’s thesis, see Werner Frank, “The History of Myth #1,” Datamation (May 1983): 252–263.

8. Brian Randell，《结语》，摘自Charles 和Ray Eames合著的《计算机视角：计算机时代的背景》，新版，I. Bernard Cohen 作序（剑桥：哈佛大学出版社，1990 年）：163。

8. Brian Randell, “Epilogue,” in Charles and Ray Eames, A Computer Perspective: Background to the Computer Age, new edition, foreword by I. Bernard Cohen (Cambridge: Harvard University Press, 1990): 163.

9.哈佛大学计算实验室，《自动序列控制计算器操作手册》，查尔斯·巴贝奇研究所重印系列，第 8 卷（剑桥：麻省理工学院出版社，1985 年；最初出版于 1946 年）。

9. Harvard University Computation Laboratory, A Manual of Operation for the Automatic Sequence Controlled Calculator, Charles Babbage Institute Reprint series, vol. 8 (Cambridge: MIT Press, 1985; originally published 1946).

10. Hopper，引自《计算机及其未来：在世界计算机先驱大会上的演讲》，威尔士拉迪德诺，1970 年 7 月，（拉迪德诺：Richard Williams and Partners）：7=3–7=4。

10. Hopper, quoted in “Computers and Their Future: Speeches Given at the World Computer Pioneer Conference,” Lladudno, Wales, July 1970, (Lladudno: Richard Williams and Partners): 7=3–7=4.

11. Grace Hopper，《ASCC 日志》，1994 年 4 月 7 日至 8 月 31 日；史密森学会档案馆；有关 Baker 问题的描述，请参阅 Herbert Grosch 的《计算机：生活中的片段》（加利福尼亚州诺瓦托：Third Millennium Books，1991 年）：51-52。

11. Grace Hopper, Log Book for the ASCC, April 7–Aug. 31, 1994; Smithsonian Institution Archives; a description of Baker’s problem is found in Herbert Grosch, Computer: Bit Slices from a Life (Novato, CA: Third Millennium Books, 1991): 51–52.

12.其中一些概念早于数字时代，例如可以追溯到麻省理工学院的微分分析仪，这是一台通过磁带序列进行编程的模拟机器。请参阅 1970 年 10 月 29 日史密森尼计算机历史项目对佩里·克劳福德的采访。

12. Some of these concepts predate the digital era and can be traced to, for example, the MIT Differential Analyzer, an analog machine that was programmed by sequences of tapes. See interview with Perry Crawford, Smithsonian Computer History Project, 29 October 1970.

13. Konrad Zuse，《规划设备》，1944 年；波恩，数学和数据处理协会，Zuse 档案，文件夹 101=024。

13. Konrad Zuse, “Planfertigungsgera¨te,” 1944; Bonn, Gesellschaft fu¨r Mathematik und Datenverarbeitung, Zuse Archive, folder 101=024.

14. Konrad Zuse，《Der Programmator》，Zeitschrift fu¨ r Angewandte Mathematik und Mechanik 32（1952）：246； Heinz Rutishauser，“Rechenplanfertigung bei programmgesteuerten Rechenmaschinen”，来自ETH 应用数学研究所的论文，no. 3、1952；另见 FL Bauer 的《Zuse 与 Rutishauser 之间：中欧数字计算的早期发展》，收录于 N. Metropolis、J. Howlett 和 Gian-Carlo Rota 编的《二十世纪计算史》（纽约）：Academic Press，1980）：505–524。

14. Konrad Zuse, “Der Programmator,” Zeitschrift fu¨ r Angewandte Mathematik und Mechanik 32 (1952): 246; Heinz Rutishauser, “Rechenplanfertigung bei programmgesteuerten Rechenmaschinen,” Mitteilungen aus dem Institut fu¨r angewandte Mathematik der ETH, no. 3, 1952; also F. L. Bauer, “Between Zuse and Rutishauser: the Early Development of Digital Computing in Central Europe,” in N. Metropolis, J. Howlett, and Gian-Carlo Rota, eds., A History of Computing in the Twentieth Century (New York: Academic Press, 1980): 505–524.

15. Rutishauser，“自动控制规划在程序设计机械中的作用”，ZAMP 3（1952）：312–313（作者译文）。

15. Rutishauser, “Automatische Rechenplanfertigung bei programmgesteuerten Rechenmaschinen,” ZAMP 3 (1952): 312–313 (author’s translation).

16. Martin Campbell-Kelly，《EDSAC 编程：剑桥大学早期的编程活动》，《计算机史年鉴》第 2 卷（1980 年）：7-36 页；另请参阅 Maurice Wilkes、DJ Wheeler 和 Stanley Gill 合著的《电子数字计算机程序的准备》（剑桥：Addison-Wesley，1951 年）。

16. Martin Campbell-Kelly, “Programming the EDSAC: Early Programming Activity at the University of Cambridge,” Annals of the History of Computing 2 (1980): 7–36; Also Maurice Wilkes, D. J. Wheeler, and Stanley Gill, The Preparation of Programs for an Electronic Digital Computer (Cambridge: Addison-Wesley, 1951).

17. Maurice Wilkes，《计算机先驱回忆录》（剑桥：麻省理工学院出版社，1985 年）：142-144。

17. Maurice Wilkes, Memoirs of a Computer Pioneer (Cambridge: MIT Press, 1985): 142–144.

18.史密森学会，NMAH 档案，Grace Hopper 论文，盒 5，文件夹 #1，ACM 会议计划，橡树岭，1949 年 4 月。

18. Smithsonian Institution, NMAH Archives, Grace Hopper Papers, Box 5, folder #1, program of ACM Meeting, Oak Ridge, April 1949.

19. JW Mauchly，致 EMCC 执行委员会的备忘录，1948 年 8 月 16 日；哈格利博物馆，编号 1825，系列 I，盒子 3：Mauchly，John；另请参阅 Richard Pearson 的《海军上将霍珀去世；计算机先驱》，《华盛顿邮报》（1992 年 1 月 2 日）。

19. J. W. Mauchly, memorandum to EMCC executive committee, 16 August 1948; Hagley Museum, Accession 1825, Series I, Box 3: Mauchly, John; also Richard Pearson, “Admiral Hopper Dies; Pioneer in Computers,” Washington Post (January 2, 1992).

20. Grace Hopper，《编译例程》，内部备忘录，埃克特-莫奇利计算机公司，费城，1953 年 12 月 31 日；史密森尼档案馆，Hopper 论文第 6 盒，第 9 个文件夹。

20. Grace Hopper, “Compiling Routines,” internal memorandum, Eckert-Mauchly Computer Corporation, Philadelphia, 31 December 1953; Box 6, folder 9, Hopper Papers, Smithsonian Archives.

21.同上。

21. Ibid.

22. Hopper 论文，NMAH，盒 5，文件夹 7。在这里，她使用“生成器”一词的意义与今天人们使用“编译器”的意义相同。

22. Hopper papers, NMAH, Box 5, folder 7. Here she uses the term “generator” in a sense that one might use “compiler” today.

23.有关“编译器”和“解释器”这两个术语最初使用的简明定义，请参阅 Grace M. Hopper 的《编译程序》，《计算机与自动化》第 2 卷（1953 年 5 月）：1-5。用 Hopper 的话来说：“使用子程序的解释方法包括固定子程序在计算机内存中的位置，并使主程序解释所谓的“伪代码”，从而引用子程序并执行它……使用子程序的编译方法包括将子程序复制到主程序中，根据需要调整内存位置以在程序中正确定位子程序并提供参数和结果”（第 2 页）。

23. For a concise definition of the terms “compiler” and “interpreter” as they were initially used, see Grace M. Hopper, “Compiling Routines,” Computers and Automation 2 (May 1953): 1–5. In Hopper’s words: “The interpretive method of using subroutines consists of fixing the location of the subroutine in the computer memory, and causing the main program to interpret what may be called a “pseudo-code,” and thus refer to the subroutine and perform it ... . The compiling method of using subroutines consists of copying the subroutine in to the main routine, adjusting memory locations as necessary to position the subroutine properly in the program and to supply arguments and results” (p. 2).

24. Wilkes、Wheeler 和 Gill 讨论了一种他们称之为“汇编子程序”的程序，该程序的功能相同。他们还描述了一种类似的程序，他们称之为“解释性”程序：参见 Wilkes、Wheeler 和 Gill 的《电子数字计算机程序的准备》，第 26-37 页。这些术语一直沿用至今，但含义不同。可以说，在 20 世纪 50 年代初，人们已经清楚地认识到，可以指示计算机接管编写程序的许多杂务，但具体接管多少以及以何种形式接管，尚不确定。另请参阅 Martin Campbell-Kelly 的《EDSAC 编程：剑桥大学早期编程活动》，《计算机史年鉴》，第 2 期（1980 年）：第 7-36 页。

24. Wilkes, Wheeler, and Gill discuss a program they call an “assembly subroutine,” which does the same thing. They also describe something similar that they call “interpretive” routines: See Wilkes, Wheeler, and Gill, The Preparation of Programs for an Electronic Digital Computer, 26–37. These terms survive to the present day but with different meanings. Suffice it to say that in the early 1950s it had become clear that the computer could be instructed to take over many of the chores of producing programs, but just how much, and in what form, was less certain. See Also Martin Campbell-Kelly, “Programming the EDSAC: Early Programming Activity at the University of Cambridge,” Annals of the History of Computing, 2 (1980): 7–36.

25. Michael Mahoney，《软件与装配线》，论文发表于电气信息技术技术史研讨会，德意志博物馆，慕尼黑，1990 年 12 月 15-19 日。

25. Michael Mahoney, “Software and the Assembly Line,” paper presented at Workshop on Technohistory of Electrical Information Technology, Deutsches Museum, Munich, December 15–19, 1990.

26. JH Laning 和 N. Zierler，《旋风 I 数学方程式翻译程序》，（马萨诸塞州剑桥：麻省理工学院工程备忘录第 E-364 号，1954 年 1 月）。国家航空航天博物馆，NBS 藏品，第 39 盒，第 8 文件夹。

26. J. H. Laning and N. Zierler, “A Program for Translation of Mathematical Equations for Whirlwind I,” (Cambridge, MA: MIT Engineering Memorandum no. E-364, January 1954). National Air and Space Museum, NBS Collection, Box 39, Folder 8.

27.参见 John Backus，《20 世纪 50 年代美国的编程——一些个人回忆》，Metropolis、Howlett 和 Rota 编，《 20 世纪计算史》，125–135 页，尤其是第 133–134 页。

27. See John Backus, “Programming in America in the 1950s—Some Personal Recollections,” in Metropolis, Howlett, and Rota, eds. History of Computing in the Twentieth Century, 125–135, especially pp. 133–134.

28. Laning 和 Zierler，《旋风 I 数学方程式翻译程序》（1954 年），卷首插图。在我查阅的一份来自美国国家标准局图书馆的副本中，有人（可能是 Samuel Alexander）划掉了“解释性”一词，并在其上方用铅笔写上“翻译”。

28. Laning and Zierler, “A Program for Translation of Mathematical Equations For Whirlwind I,” (1954) frontispiece. In the copy that I examined, from the library of the National Bureau of Standards, someone—possibly Samuel Alexander—crossed out the word “interpretive” and wrote in pencil above it “translated.”

29. Backus，《美国的编程》。

29. Backus, “Programming in America.”

30. Charles W. Adams 和 JH Laning Jr.，《麻省理工学院自动编码系统：综合、暑期课程和代数》，《数字计算机自动编程研讨会》（华盛顿特区：美国海军，1954 年）：64。

30. Charles W. Adams and J. H. Laning Jr., “The MIT Systems of Automatic Coding: Comprehensive, Summer Session, and Algebraic,” in Symposium on Automatic Programming of Digital Computers (Washington, DC: U.S. Navy, 1954): 64.

31. Donald Knuth，《编程语言的早期发展》，载于Metropolis、Howlett 和 Rota 编著《20 世纪计算史》，第 239 页。

31. Donald Knuth, “The Early Development of Programming Languages,” in Metropolis, Howlett, and Rota, eds., History of Computing in the Twentieth Century, 239.

32. Paul Armer，《SHARE——对合作努力的颂扬》，《计算机历史年鉴》，2：2（1980 年 4 月）：122-129。Armer 说，“SHARE”不代表任何含义；其他人说它代表“社会帮助避免冗余努力”。

32. Paul Armer, “SHARE—a Eulogy to Cooperative Effort,” Annals of the History of Computing, 2: 2 (April 1980): 122–129. Armer says that “SHARE” did not stand for anything; others say it stood for “Society to Help Avoid Redundant Effort.”

33.Donald Knuth，《排序和搜索》（马萨诸塞州雷丁：Addison-Wesley，1973 年）：第 3 页。

33. Donald Knuth, Sorting and Searching (Reading, MA: Addison-Wesley, 1973): 3.

34.例如，请参阅 CAR Hoare 的作品。另请参阅 Knuth 的《排序和搜索》。

34. See, for example, the work of C. A. R. Hoare. See also Knuth, Sorting and Searching.

35.冯·诺依曼致赫尔曼·戈德斯坦，1945 年 5 月 8 日；哈格利博物馆，编号 1825，盒子 74，系列 II，Chron 档案，1945 年 5 月 - 1945 年 10 月。

35. Von Neumann to Herman Goldstine, 8 May 1945; Hagley Museum, Accession 1825, Box 74, series II, Chron File, May 1945–October 1945.

36. Donald Knuth，《冯·诺依曼的第一个计算机程序》，《计算概览》 2（1970 年 12 月）：247-260。

36. Donald Knuth, “Von Neumann’s First Computer Program,” Computing Surveys 2 (December 1970): 247–260.

37. 1947 年至 1950 年 6 月，霍尔伯顿使用姓氏斯奈德。参见对霍尔伯顿的采访，UNIVAC 口述历史会议，史密森学会，1990 年 5 月 17-18 日，第 52 页；史密森学会档案。雷明顿·兰德公司收购 EMCC 后不久，她离开前往华盛顿特区外的美国海军戴维·泰勒模型盆地，继续担任 UNIVAC #6 的程序员。

37. Between 1947 to June of 1950 Holberton went by the surname Snyder. See interview with Holberton, UNIVAC Oral History Conference, Smithsonian Institution, May 17–18, 1990, 52; Smithsonian Institution Archives. Soon after Remington Rand acquired EMCC, she left for the U.S. Navy’s David Taylor Model Basin just outside Washington, D.C., where she continued work as a programmer on UNIVAC #6.

38. Knuth，《排序和搜索》，第 386 页。Holberton 的例程在美国海军数学顾问小组的数字计算机自动编程研讨会（华盛顿特区，1954 年）：第 34-39 页中有描述。

38. Knuth, Sorting and Searching, 386. Holberton’s routine is described in U.S. Naval Mathematics Advisory Panel, Symposium on Automatic Programming for Digital Computers (Washington, DC, 1954): 34–39.

39. Eric Weiss 编辑，《计算机使用—基础知识》（纽约：McGraw-Hill，1969 年）。

39. Eric Weiss, ed., Computer Usage—Fundamentals (New York: McGraw-Hill, 1969).

40. Holberton，UNIVAC 口述历史会议访谈，史密森学会，1990 年 5 月 17-18 日，第 52 页；史密森学会档案馆。

40. Holberton, interview in UNIVAC oral history conference, Smithsonian Institution, May 17–18, 1990, p. 52; Smithsonian Institution Archives.

41.最新版本于 1992 年被国际标准组织接受，称为“Fortran 90”（现在拼写为专有名词，不再是首字母缩略词）。Fortran 的一个版本针对并行、非冯·诺依曼架构计算机，也正在开发中；它目前被称为高性能 Fortran。请参阅 Ian Stockdale 的“供应商、国际标准社区采用 Fortran 90”，NAS 新闻（NASA 艾姆斯研究中心，1992 年 9 月）：1-8。

41. The most recent version, accepted by the International Standards Organization in 1992, is called “Fortran 90,” (now spelled as a proper noun, no longer as an acronym). A version of Fortran aimed at parallel, non-von Neumann architecture computers, is also under development; it is presently called High Performance Fortran. See Ian Stockdale, “Vendors, International Standards Community Adopt Fortran 90,” NAS News (NASA Ames Research Center, September 1992): 1–8.

42. Backus，《编程在美国》。ALGOL 语言的开发者之一、慕尼黑工业大学的 FL Bauer 教授向作者指出，许多 ALGOL 编译器也很快、很高效，因此，在他看来，这显然不是 FORTRAN 语言成功的唯一原因。

42. Backus, “Programming in America.” Professor F. L. Bauer of the Technical University of Munich, who was one of the developers of ALGOL, pointed out to the author that many ALGOL compilers were also fast and efficient, so this, in his view, was obviously not the sole reason for FORTRAN’s success.

43. Backus，《美国的编程》，第131页。

43. Backus, “Programming in America,” 131.

44.该委员会被称为“CODASYL”：数据系统语言会议。它不是通常意义上的委员会；COBOL 的实际工作是由另外两个委员会完成的，其中包括计算机制造商和政府机构的代表。请参阅 Jean Sammett 的《编程语言：历史和基础》（新泽西州恩格尔伍德克利夫斯：Prentice Hall，1969 年），第 V.3 节。

44. The committee was called “CODASYL”: Conference On DAta SYstems Languages. It was not a committee in the normal sense of the word; actual work on COBOL was done by two other committees that included representatives of computer manufacturers and government agencies. See Jean Sammett, Programming Languages: History and Fundamentals (Englewood Cliffs, NJ: Prentice Hall, 1969), section V.3.

45. “一个编译器即将问世！” Datamation（1959 年 5 月 = 6 月）：15；另请参阅 Saul Rosen，“编程系统和语言：历史概览”，Proc. SJCC 25（1964）：1-15。

45. “One Compiler, Coming Up!” Datamation (May=June 1959): 15; also Saul Rosen, “Programming Systems and Languages: a Historical Survey,” Proc. SJCC 25 (1964): 1–15.

46.该示例取自 Sammett，《编程语言》，337 页。

46. The example is taken from Sammett, Programming Languages, 337.

47.这部电影于 1968 年上映。在后来的一个场景中，HAL 表示他是在伊利诺伊州厄巴纳诞生的，1968 年，这里是雄心勃勃的超级计算机项目“Illiac-IV”的所在地。

47. The movie was released in 1968. In a later scene, HAL states that he was created in Urbana, Illinois, which in 1968 was the location of an ambitious supercomputer project known as “Illiac-IV.”

48. Paul Ceruzzi，“欧洲计算历史方面”，论文发表于 1987 年 10 月 8-10 日在内布拉斯加州奥马哈举行的第 12 届欧洲研究会议。

48. Paul Ceruzzi, “Aspects of the History of European Computing,” Paper presented at the 12th European Studies Conference, Omaha, NE, 8–10 October, 1987.

49. Grace Hopper，《Keynotestext1 Address》，载于 Richard L. Wexelblatt 主编的《编程语言史》（纽约：Academic Press，1981 年）：7-20。

49. Grace Hopper, “Keynotestext1 Address,” in Richard L. Wexelblatt, ed., History of Programming Languages (New York: Academic Press, 1981): 7–20.

50.同上；另请参阅 Sammet，《编程语言》。

50. Ibid.; see also Sammet, Programming Languages.

51.Sammet，《编程语言》，11。

51. Sammet, Programming Languages, 11.

52. GF Ryckman, Proc WJCC 1960, 341–343；另请参阅 William Orchard-Hays，“编程系统的演进”，IRE Proc . 49（1961 年 1 月）：283–295。

52. G. F. Ryckman, Proc WJCC 1960, 341–343; also William Orchard-Hays, “The Evolution of Programming Systems,” IRE Proc. 49 (January 1961): 283–295.

53.斜线星号命令由 Larry Josephson 设计，他后来又在纽约另类广播电台 WBAI 担任地下唱片骑师。Larry Josephson 与作者的私人通信。

53. The slash-asterisk command was designed by Larry Josephson, who later had a second career as an underground disk jockey with the New York alternative radio station WBAI. Larry Josephson, private communication to the author.

54. Sammet，《编程语言》，205-215；Bob Rosin，个人通信，1995 年 6 月 23 日。

54. Sammet, Programming Languages, 205–215; Bob Rosin, personal communication, 23 June 1995.

55.稍后描述的 PL=I 编程语言也是这项统一努力的一部分；它将取代商业领域的 COBOL 和科学应用领域的 FORTRAN。

55. The PL=I programming language, described later, was also a part of this unifying effort; it was to replace COBOL for business, and FORTRAN for scientific applications.

56. Fred Brooks，《人月神话：软件工程随笔》（马萨诸塞州雷丁：Addison-Wesley，1975 年）。

56. Fred Brooks, The Mythical Man-Month: Essays on Software Engineering (Reading, MA: Addison-Wesley, 1975).

57.例如，请参阅密歇根大学工程学院，《逻辑在高级数字计算机编程中的应用》，《工程师和科学家强化课程笔记》，1957 年夏季课程。国家航空航天博物馆，NBS 收藏。

57. See, for example, University of Michigan, College of Engineering, “Applications of Logic to Advanced Digital Computer Programming,” notestext1s for an Intensive Course for the Engineer and Scientist, 1957 Summer Session. National Air and Space Museum, NBS Collection.

58. Paul Ceruzzi，“电子技术和计算机科学”，计算史年鉴，10：4（1989）：257-275；Louis Fein，“大学在计算机、数据处理和相关领域中的作用”，通信，ACM 2（1959）：7-14。

58. Paul Ceruzzi, “Electronics Technology and Computer Science,” Annals of the History of Computing, 10: 4 (1989): 257–275; also Louis Fein, “The Role of the University in Computers, Data Processing, and Related Fields,” Communications, ACM 2 (1959): 7–14.

59.斯坦福大学档案馆，乔治·福赛斯论文；另见唐纳德·E·克努斯的《乔治·福赛斯与计算机科学的发展》，《通讯》，ACM 15：8（1972）：721-726。

59. Stanford University Archives, George Forsythe papers; also Donald E. Knuth, “George Forsythe and the Development of Computer Science,” Communications, ACM 15: 8 (1972): 721–726.

60. Newell、Perlis 和 Simon，致编辑的信，Science 157（1967 年 9 月 22 日）：1373-1374。

60. Newell, Perlis, and Simon, letter to the editor, Science 157 (22 September, 1967): 1373–1374.

61. Herbert Simon，《人工智能科学》，第 2 版（剑桥：麻省理工学院出版社，1981 年）。

61. Herbert Simon, The Sciences of the Artificial, 2nd edition (Cambridge: MIT Press, 1981).

62. Newell、Perlis 和 Simon，致编辑的信，上文引文，第 1374 页。

62. Newell, Perlis, and Simon, letter to the editor, op. cit., p. 1374.

63. 通信，ACM 11 (1968 年 3 月)：147。

63. Communications, ACM 11 (March 1968): 147.

64. Seymour V. Pollack，“计算机科学的发展”，载 Seymour Pollack 主编，《计算机科学研究》（华盛顿特区：美国数学协会，1982 年）：1-51。

64. Seymour V. Pollack, “The Development of Computer Science,” in Seymour Pollack, ed., Studies in Computer Science (Washington, DC: Mathematical Association of America, 1982): 1–51.

65.同上，第35页。

65. Ibid., 35.

66. Donald E. Knuth，《计算机编程艺术》，第 1 卷：基本算法（马萨诸塞州雷丁：Addison-Wesley，1968 年）。Knuth 于 1969 年出版了第 2 卷，1973 年出版了第 3 卷；到目前为止，他还没有完成剩下的四卷。在最近的一次采访中，他表示希望能够在 2003 年出版第 4 卷（分为三部分），并在 2008 年出版第 5 卷。

66. Donald E. Knuth, The Art of Computer Programming, vol. 1: Fundamental Algorithms (Reading, MA: Addison-Wesley, 1968). Knuth published volume 2 in 1969 and volume 3 in 1973; he has thus far not produced the remaining four volumes. In a recent interview he expressed a hope that he would be able to publish volume 4 (in three parts) in 2003, and volume 5 in 2008.

67. 通信，ACM 11 (1968 年 3 月)：147。

67. Communications, ACM 11 (March 1968): 147.

68. “商标语言保护不明智”，Communications，ACM 11：3（1968 年 3 月）：148-149。

68. “Language Protection by Trademark Ill-advised,” Communications, ACM 11: 3 (March 1968): 148–149.

69.据 Mooers 在接受作者采访时称，TRAC 语言被“窃取”并以“MINT”的名义无限制发布——“Mint Is Not TRAC”。

69. According to Mooers, in an interview with the author, the TRAC language was “stolen” and issued, without restrictions, under the name “MINT”—“Mint Is Not TRAC.”

70.比尔·盖茨是同意穆尔斯观点的人之一，他写信给穆尔斯，表达了对穆尔斯的同情，并表示有兴趣追随穆尔斯的行动。这封信现在被保存在查尔斯·巴贝奇研究所的穆尔斯文件中。

70. One person who agreed with Mooers was Bill Gates, who wrote to him expressing sympathy and an interest in following Mooer’s crusade. The letter is now among Mooers’s papers at the Charles Babbage Institute.

71. Peter Naur 和 Brian Randell，“软件工程；北约科学委员会主办的会议报告”，1968 年 10 月 7-11 日（德国加米施：北约，1969 年）。

71. Peter Naur and Brian Randell, “Software Engineering; Report on a Conference Sponsored by the NATO Science Committee,” 7–11 October 1968 (Garmisch, Germany: NATO, 1969).

72. James Tomayko，“二十年回顾：北约软件工程会议”，计算机史年鉴，11：2（1989）：132-143。

72. James Tomayko, “Twenty Year Retrospective: the NATO Software Engineering Conferences,” Annals of the History of Computing, 11: 2 (1989): 132–143.

73. Thomas J. Bergin 和 James (Jay) Horning，“达格斯图尔会议”，计算史年鉴19:3（1997）：74-76。

73. Thomas J. Bergin and James (Jay) Horning, “Dagstuhl Conference,” Annals of the History of Computing 19: 3 (1997): 74–76.

74. Franklin Fisher、James W. McKie、Richard B. Mancke，《IBM 和美国数据处理行业》（纽约：Praeger，1983 年）：176-177。

74. Franklin Fisher, James W. McKie, Richard B. Mancke, IBM and the U.S. Data Processing Industry (New York: Praeger, 1983): 176–177.

75. David C. Mounce，CICS：一部轻松编年史（Hursely，英国：IBM，1994 年）：i。

75. David C. Mounce, CICS: a Light Hearted Chronicle (Hursely, UK: IBM, 1994): i.

76. Peter Salus，《UNIX 的四分之一世纪》（马萨诸塞州雷丁：Addison-Wesley，1994 年）；后面将更全面地讨论 UNIX。

76. Peter Salus, A Quarter Century of UNIX (Reading, MA: Addison-Wesley, 1994); UNIX will be discussed more fully later on.

77. Dennis M. Ritchie，《C 语言的发展》，载 Thomas J. Bergin Jr. 和 Richard G. Gibson Jr. 编，《编程语言史—II》（纽约：ACM Press，1996 年）：671-698。

77. Dennis M. Ritchie, “The Development of the C Programming Language,” in Thomas J. Bergin Jr. and Richard G. Gibson Jr., eds., History of Programming Languages—II (New York: ACM Press, 1996): 671–698.

78.同上，第673页。

78. Ibid., 673.

79. Rosen，《编程系统和语言：历史调查》，第 12 页。

79. Rosen, “Programming Systems and Languages: a Historical Survey,” 12.

80. CH Lindsey，“ALGOL-68 的历史”，ACM Sigplan Notices 28：3（1993 年 3 月）：97-132。

80. C. H. Lindsey, “A History of ALGOL-68,” ACM Sigplan Notices 28: 3 (March 1993): 97–132.

81.例如，委员会成员之间曾发生过激烈的争论，争论的焦点是 ALGOL-68 应如何处理用逗号分隔千位、用点分隔小数位的情况，或者像在德国常见的那样用逗号分隔千位、用点分隔小数位；例如，2,555.32 在德国写为 2.555,32。

81. For example, there was an acrimonious debate among committee members over how ALGOL-68 should handle the use of a comma to separate thousands and a point for the decimal, or the opposite, as is common in Germany; for example, 2,555.32 in Germany is written 2.555,32.

82. Nicholas Wirth，“关于 PASCAL 开发的回忆”，ACM Sigplan Notices 28：3（1993 年 3 月）：333-342。

82. Nicholas Wirth, “Recollections about the Development of PASCAL,” ACM Sigplan Notices 28: 3 (March 1993): 333–342.

83.兰德公司数学家梅里尔·弗勒德声称，他在 20 世纪 40 年代末的一份备忘录中创造了这个术语，但显然这个术语在十年后被计算机界重新使用。弗勒德，《Datamation》杂志致编辑的信（1984 年 12 月 1 日）：15-16。弗勒德说，他是间接听到艾克的评论的。“软件”一词直到 1959 年左右才出现在印刷品中，当时该术语逐渐开始出现在计算机期刊和行业媒体中。

83. Merrill Flood, a mathematician with the RAND Corporation, claimed that he coined the term in a memorandum written in the late 1940s, but apparently the term was re-created among computer people ten years later. Flood, letter to the editor, Datamation (December 1, 1984): 15–16. Flood says that he heard Ike’s comment secondhand. “Software” did not appear in print until around 1959, when the term gradually began appearing in the computer journals and trade press.

第四章

Chapter 4

1. “美国国税局服务中心的 ADP 史”，录像带，美国国税局德克萨斯州奥斯汀服务中心，1991 年 12 月 11 日。

1. “History of ADP in IRS Service Centers,” videotape, IRS Austin, Texas, Service Center, December 11, 1991.

2.同上。

2. Ibid.

3.美国国税局，《国税局的数据处理历史和演变》，作者保存未注明日期的打字稿。

3. U.S. Internal Revenue Service, “Data Processing History and Evolution in IRS,” undated typescript in author’s possession.

4.数字设备公司，《PDP-8 排版系统》手册，约 1966 年，作者所有；另见 Eric Weiss 编辑的《计算机使用——应用》（纽约：计算机使用公司，1969 年），第 6 章。

4. Digital Equipment Corporation, “PDP-8 Typesetting System,” brochure, ca. 1966, in author’s possession; also Eric Weiss, ed., Computer Usage—Applications (New York: Computer Usage Corporation, 1969), chapter 6.

5. Gerald Brock，《美国计算机产业：市场力量研究》（马萨诸塞州剑桥：Ballinger，1975 年）。

5. Gerald Brock, The U.S. Computer Industry: A Study in Market Power (Cambridge, MA: Ballinger, 1975).

6. IBM 的研究实验室是软盘存储、FORTRAN 和关系数据库等创新的发源地。

6. IBM’s research labs were the origins of floppy disk storage, FORTRAN, and relational databases, among other innovations.

7. Samuel Snyder，“美国密码组织对数字计算机产业的影响”，系统与软件杂志1 (1979): 87-102。

7. Samuel Snyder, “Influence of U.S. Cryptologic Organizations on the Digital Computer Industry,” Journal of Systems and Software 1 (1979): 87–102.

8. Saul Rosen，《电子计算机：历史调查》，《计算调查》 1（1969 年 3 月）：7-36。

8. Saul Rosen, “Electronic Computers: a Historical Survey, Computing Surveys 1 (March 1969): 7–36.

9.布罗克，《美国计算机产业》。

9. Brock, The U.S. Computer Industry.

10. Paul Forman，“量子电子学背后：1940-1960 年代国家安全作为美国物理研究的基础”，HSPS 18:1（1987）：149-229。

10. Paul Forman, “Behind Quantum Electronics: National Security as Basis of Physical Research in the United States, 1940–1960,” HSPS 18: 1 (1987): 149–229.

11. George Ballester，《马萨诸塞州蓝十字蓝盾计算机简史》，未发表论文，约 1984 年，作者所有；另请参阅 Charles J. Bashe 等人的《IBM 的早期计算机》（剑桥：麻省理工学院出版社，1986 年）：50、115。

11. George Ballester, “A Brief History of Computers at Massachusetts Blue Cross=Blue Shield,” unpublished paper, ca. 1984, in author’s possession; also Charles J. Bashe et al., IBM’s Early Computers (Cambridge: MIT Press, 1986): 50, 115.

12. Ballester，“马萨诸塞州蓝十字蓝盾的计算机简史”。

12. Ballester, “A Brief History of Computers at Massachusetts Blue Cross=Blue Shield.”

13.同上。

13. Ibid.

14.同上。

14. Ibid.

15.美国国家航空航天局艾姆斯研究中心，《更换 IBM 7040=7094 直接耦合系统必要性的论证》，加利福尼亚州莫菲特场（1967 年 3 月 31 日）：第 8 页。美国国家航空航天博物馆档案馆。

15. NASA, Ames Research Center, “A Justification of the Need to Replace the IBM 7040=7094 Direct Couple System,” Moffett Field, California (March 31, 1967): 8. National Air and Space Museum Archives.

16.同上；另见“计算部门员工会议记录”，1968 年 2 月 13 日，NASA-Ames 研究中心。NASM 档案。

16. Ibid.; also “Computation Division Staff Meeting Minutes,” February 13, 1968, NASA-Ames Research Center. NASM Archives.

17.美国国家航空航天局艾姆斯研究中心，《ADPE 采购计划——A 类》，1967 年 9 月 25 日备忘录；另请参阅 Bill Mead，《计算部门联系报告，1969 年 6 月 20 日》，美国国家航空航天局档案馆。

17. NASA-Ames Research Center, “ADPE Acquisition Plan—category A,” memorandum September 25, 1967; also Bill Mead, “Computation Division Contact Report, June 20, 1969,” NASM Archives.

18.美国国家航空航天局艾姆斯研究中心，“IBM 1401 系统”，通信文件，NASM 档案。

18. NASA-Ames Research Center, “IBM 1401 System,” correspondence file, NASM Archives.

19.美国宇航局艾姆斯研究中心，《中央计算机设施 RFP 评估委员会调查结果》，1969 年 9 月 15 日备忘录；美国宇航局艾姆斯研究中心文件，文件夹 SYS-AA：IBM 360=67 系统，FY-69-70，NASM 档案。

19. NASA-Ames Research Center, “Findings of the Evaluation Committee on the RFP of the Central Computer Facility,” memorandum of 9=15=69; NASA-Ames file, Folder SYS-AA: IBM 360=67 SYSTEM, FY-69-70, NASM Archives.

20. Emerson W. Pugh、Lyle R. Johnson 和 John H. Palmer，《IBM 的 360 和早期 370 系统》（剑桥：麻省理工学院出版社，1991 年）：630-631。

20. Emerson W. Pugh, Lyle R. Johnson, and John H. Palmer, IBM’s 360 and Early 370 Systems (Cambridge: MIT Press, 1991): 630–631.

21.美国宇航局艾姆斯研究中心，韦恩·哈撒韦致计算机系统分部负责人的备忘录，1971 年 3 月 31 日；文件夹 SYS-AA：IBM 360=67（ARPA 网络），NASM 档案。

21. NASA-Ames Research Center, Memorandum, Wayne Hathaway to Chief, Computer Systems Branch, March 31, 1971; folder SYS-AA: IBM 360=67 (ARPA Network), NASM Archives.

22.美国国家航空航天局艾姆斯研究中心，文件夹编号 24：IBM 360=50 计算机，A 类，NASM 档案。

22. NASA-Ames Research Center, folder #24: IBM 360=50 Computer, Category A, NASM Archives.

23.以下叙述的材料来自美国国税局提供给作者的大量材料，包括一部影片《国税局服务中心的 ADP 历史》，德克萨斯州奥斯汀国税局服务中心，1991 年 12 月 11 日；Daniel Capozzoli 的《数据处理的早期》，《计算机服务通讯》，美国国税局，华盛顿特区，1987 年 7 月；以及《ADP-历史》，未分页、未注明日期的打字稿（约 1970 年），美国国税局，华盛顿特区。作者谨此感谢美国国税局历史学家 Shelly Davis 的支持和帮助。

23. Material for the following account has been drawn from a number of materials supplied to the author from the IRS, including a film, “History of ADP in IRS Service Centers,” IRS Austin, Texas, Service Center, December 11, 1991; Daniel Capozzoli, “The Early Years of Data Processing,” Computer Services Newsletter, US Internal Revenue Service, Washington, DC, July 1987; and “ADP-History,” unpaginated, undated typescript (ca. 1970), IRS, Washington, DC. The author wishes to acknowledge the support of Shelly Davis, IRS historian, for assistance.

24. Capozzoli，“数据处理的早期阶段”，1。

24. Capozzoli, “The Early Years of Data Processing,” 1.

25.同上。

25. Ibid.

26.为了进行此比较，我假设每 2,400 英尺长的磁带容量约为 5 兆字节，一个手提箱装有 10 盘磁带，飞行时间为 4 小时。请参阅 Bashe 等人的《IBM 的早期计算机》（剑桥：麻省理工学院出版社，1986 年）第 215 页；Pugh 等人的《IBM 的 360 和早期 370 系统》，第 530 页。

26. For this comparison I am assuming a tape capacity of about 5 megabytes per 2,400-foot tape, a suitcase containing ten tapes, and a four-hour flight. See Bashe et al., IBM’s Early Computers (Cambridge: MIT Press, 1986) 215; Pugh et al., IBM’s 360 and Early 370 Systems, 530.

27.美国国税局，“自动数据处理的历史”。

27. Internal Revenue Service, “History of Automatic Data Processing.”

28.美国审计总署，《保护纳税人信息——对拟议的计算机化税务管理系统的评估》，美国审计长提交给国会的报告，LCD-76-115 [1977 年 1 月 17 日]。

28. U.S. General Accounting Office, “Safeguarding Taxpayer Information—An Evaluation of the Proposed Computerized Tax Administration System,” United States Comptroller General, Report to the Congress, LCD-76-115 [January 17, 1977].

29.同上，第4页。

29. Ibid., 4.

30.同上。另请参阅《Computerworld》11:8（1977 年 2 月 21 日）：1、6。

30. Ibid. See also Computerworld 11: 8 (February 21, 1977): 1, 6.

31.同上。

31. Ibid.

32.美国参议院，第 95 届国会，《1978 财年财政、邮政服务和一般政府拨款听证会，第 1 部分》；美国众议院，《拨款小组委员会听证会》，财政部财政、邮政服务和一般政府拨款小组委员会，第 1 部分，美国国家档案馆。引文来自众议院举行的听证会，由俄亥俄州众议员克拉伦斯·米勒提出。

32. U.S. Senate, 95th Congress, Hearings on Treasury, Postal Service, and General Government Appropriations, FY 1978, Part 1; U.S. House of Representatives, Hearings Before a Subcommittee of Appropriations, Subcommittee on the Treasury, Postal Service, and General Government Appropriations, Part 1, Department of the Treasury, U.S. National Archives. The quotations are from the hearing held in the House, and were posed by Rep. Clarence Miller of Ohio.

33.美国国税局历史办公室，私人通信；另见美国参议院拨款委员会，美国财政部、美国邮政服务和一般政府拨款小组委员会，第 95 届国会听证会，关于 HR 12930，1978 年 3 月 7 日，第 2 部分，第 269-297 页。

33. IRS History Office, private communication; also U.S. Senate, Committee on Appropriation, Subcommittee on the Department of the U.S. Treasury, U.S. Postal Service, and General Government Appropriations, Hearings before the 95th Congress, on H.R. 12930, March 7, 1978, part 2, 269–297.

34.美国众议院，第 95 届国会拨款委员会小组委员会听证会，1978 年 3 月 12 日，第 438 页。

34. U.S. House of Representatives, Hearings before a Subcommittee of the Committee on Appropriations, 95th Congress, March 12, 1978, 438.

35.整个故事在戴维·伯纳姆的《计算机国家的崛起》（纽约：兰登书屋，1983 年）中有详细描述。

35. The whole story is well told in David Burnham, The Rise of the Computer State (New York: Random House, 1983).

36.在本讨论中，我强调了标准商用大型计算机上顺序批处理操作的替代方案。这并不意味着实时（NASA-Houston）、在线（拟议的 IRS）和分时（NASA-Ames）用途都是一样的。它们完全不同，但它们都偏离了那个时代的真正计算模型，而那个时代是前计算机制表机时代的产物。

36. In this discussion I emphasize alternatives to sequential batch operation on standard commercial mainframe computers. That does not mean that real-time (NASA-Houston), on-line (proposed IRS), and time-shared (NASA-Ames) uses are all the same. They are quite different, but they all depart from the true model of computing of that era, which was a descendent of the precomputer tabulator era.

37. James Tomayko，《航天中的计算机：NASA 经验》（华盛顿特区：NASA，1988 年）：244；SE James，“载人航天实时计算机系统的演变”，IBM J. Res. Dev . 25（181）：418。

37. James Tomayko, Computers in Spaceflight: the NASA Experience (Washington, DC: NASA, 1988): 244; S. E. James, “Evolution of Real-time Computer Systems for Manned Spaceflight,” IBM J. Res. Dev. 25 (181): 418.

38. Paul Ceruzzi，《超越极限：飞行进入计算机时代》（剑桥：麻省理工学院出版社，1989 年），第 9 章。

38. Paul Ceruzzi, Beyond the Limits: Flight Enters the Computer Age (Cambridge: MIT Press, 1989), chapter 9.

39. SE James，“载人航天实时计算机系统的演进”，IBM J Res Dev 25（1981）：245；另请参阅 Saul I. Gass，“水星计划实时计算与数据流系统”，Proc. Eastern JCC（华盛顿特区）：33-46。

39. S. E. James, “Evolution of Real Time Computer Systems for Manned Spaceflight,” IBM J Res Dev 25 (1981): 245; also Saul I. Gass, “Project Mercury Real-Time Computational and Data-flow System,” Proc. Eastern JCC (Washington, DC): 33–46.

40. Gass，《水星计划》，第37页。

40. Gass, “Project Mercury,” 37.

41. Marilyn Scott 和 Robert Hoffman，《水星编程系统》，《东部 JCC 20 报》（1961 年 12 月 12-14 日）：第 47-53 页。小型计算机设计人员独立开发了此方案的变体，以消除特殊陷阱处理器。此后，它被称为“优先级驱动中断”，是各种规模的现代计算系统的基石。

41. Marilyn Scott, and Robert Hoffman, “The Mercury Programming System,” Proc. Eastern JCC 20 (December 12–14, 1961): 47–53. A variation of this scheme was independently developed by minicomputer designers to eliminate the special trap processor. It has since become known as a “priority-driven interrupt” and is a cornerstone of modern computing systems of all sizes.

42. Tomayko，《太空飞行中的计算机》，246。

42. Tomayko, Computers in Spaceflight, 246.

43. Bashe，IBM 的早期计算机。IBM 区分了“在线”和“实时”操作，前者处理数据的速度不必与输入系统的数据速度相同或更快。航空预订系统是“在线”的：用户希望系统快速响应，但愿意忍受短暂的延迟，最多几秒钟。这种延迟，无论多么短暂，对于实时系统（例如控制导弹发射的系统）来说都是致命的。

43. Bashe, IBM’s early Computers. IBM distinguished between “on-line” and “realtime” operation, in that the former did not have to process data at the same or faster rate than data were being fed into the system. An airline reservation system is “on-line”: a user wants the system to respond quickly but is willing to tolerate brief delays, up to a few seconds. Such delays, however brief, would be fatal to a real-time system, such as the one controlling a missile launch.

44. Mark Mergen（IBM 公司），私人通信，1994 年 7 月 8 日。我还要感谢 IBM 的 David Larkin 帮助我追查这一历史线索。“SPOOL”代表“在线同步外围操作”，即在处理器执行其他工作时进行打印的能力。

44. Mark Mergen (IBM Corporation), private communication, July 8, 1994. I also want to thank David Larkin of IBM for assisting me with tracking down this thread of history. ‘SPOOL’ stood for “simultaneous peripheral operations on line,” as in the ability to print while the processor was doing another job.

45.这种情况类似于美国的电话系统，在 20 世纪 80 年代初 AT&T 解散之前，该系统是受监管的垄断企业。住宅用户通常租用电话，而不是拥有电话，并且禁止他们对家中的仪器或线路进行任何改装。

45. The situation was analogous to the U.S. phone system, which before the breakup of AT&T in the early 1980s was a regulated monopoly. Residential customers typically rented, not owned, their telephones, and they were forbidden to make any modifications to the instruments or to the wiring in their homes.

46. David E. Lundstrom，《UNIVAC 的几个好人》（剑桥：麻省理工学院出版社，1987 年）：136；另见 David K. Allison，《二战以来美国海军的研究与发展》，收录于 Merritt Roe Smith 主编的《军事企业与技术变革》（剑桥：麻省理工学院出版社，1985 年）：289-328。NTDS 的字长为 30 位。

46. David E. Lundstrom, A Few Good Men from UNIVAC (Cambridge: MIT Press, 1987): 136; also David K. Allison, “U.S. Navy Research and Development Since World War II,” in Merritt Roe Smith, ed., Military Enterprise and Technological Change (Cambridge: MIT Press, 1985): 289–328. The NTDS had a 30-bit word length.

47.Lundstrom，《好人寥寥》，67页。

47. Lundstrom, A Few Good Men, 67.

48. Seymour Cray，接受 David K. Allison 采访（1995 年 5 月），史密森学会计算机历史项目。“小型计算机”一词具有误导性，因为它仅表示物理尺寸；但是，它已被广泛接受，因此我将在本章中继续使用它。

48. Seymour Cray, interview with David K. Allison (May 1995), Smithsonian Institution Computer History Project. The term “minicomputer” is misleading because it suggests only physical size; however, it has become so accepted that I will continue to use it in this chapter.

49. Montgomery Phister Jr.，《Quotron II：一种用于股票市场数据通信的早期多程序多处理器》，《计算机史年鉴》第 11 卷（1989 年）：109-126 页；作者收藏的纽约证券交易所未注明日期的小册子。

49. Montgomery Phister Jr., “Quotron II: An Early Multiprogrammed Multiprocessor for the Communication of Stock Market Data,” Annals of the History of Computing 11 (1989): 109–126; also New York Stock Exchange, undated brochure in the author’s collection.

50. Jamie Parker Pearson，《数字化工作》（马萨诸塞州贝德福德：Digital Press，1992）：6-9。

50. Jamie Parker Pearson, Digital at Work (Bedford, MA: Digital Press, 1992): 6–9.

51. C. Gordon Bell、J. Craig Mudge 和 John McNamara，《计算机工程：DEC 对硬件系统设计的视角》 (马萨诸塞州贝德福德：Digital Press，1978)：125–127。

51. C. Gordon Bell, J. Craig Mudge, and John McNamara, Computer Engineering: a DEC View of Hardware Systems Design (Bedford, MA: Digital Press, 1978): 125– 127.

52. “TX-0：它的过去和现在”，计算机博物馆报告 8（1984 年春季）：2-11。

52. “The TX-0: Its Past and Present,” Computer Museum Report 8 (Spring 1984): 2–11.

53.Bell等人，计算机工程，126-127。

53. Bell et al., Computer Engineering, 126–127.

54.同上，第129页。

54. Ibid., 129.

55.Fisher，IBM，34，308。

55. Fisher, IBM, 34, 308.

56. Bell 等人，《计算机工程》，136-139；Pearson，《数字化工作》，16-21。

56. Bell et al., Computer Engineering, 136–139; Pearson, Digital at Work, 16–21.

57.史蒂芬·利维，《黑客：计算机革命的英雄》（纽约：Anchor=Doubleday，1984 年）。

57. Steven Levy, Hackers: Heroes of the Computer Revolution (New York: Anchor=Doubleday, 1984).

58. Pearson，《数字化工作》，143。

58. Pearson, Digital at Work, 143.

59.格伦·里夫金、乔治·哈拉尔，《终极企业家：肯·奥尔森和数字设备公司的故事》（芝加哥：当代图书，1988 年）。

59. Glenn Rifkin, and George Harrar, The Ultimate Entrepreneur: The Story of Ken Olsen and Digital Equipment Corporation (Chicago: Contemporary Books, 1988).

60.Bell等人，计算机工程，181。

60. Bell et al., Computer Engineering, 181.

61.同上，第141页。

61. Ibid., 141.

62. A. Clark Wesley，“LINC 成立较早，规模较小”，载于 Adele Goldberg 主编的《个人工作站的历史》（纽约：ACM Press，1988 年）：第 347-391 页。

62. A. Clark Wesley, “The LINC was Early and Small,” in Adele Goldberg, ed., A History of Personal Workstations (New York: ACM Press, 1988): 347–391.

63. PR Morris，世界半导体工业史（伦敦：Peter Peregrinus，1990）：42。

63. P. R. Morris, A History of the World Semiconductor Industry (London: Peter Peregrinus, 1990): 42.

64. C. Gordon Bell 和 Allen Newell，《计算机结构：阅读材料和范例》（纽约： McGraw-Hill，1971 年）；Bashe 等人，《IBM 的早期计算机》，第 448–449 页。

64. C. Gordon Bell and Allen Newell, Computer Structures: Readings and Examples (New York: McGraw-Hill, 1971); Bashe et al., IBM’s Early Computers, 448–449.

65.数字设备公司，“PDP-8 排版系统”，广告宣传册，1966 年，马萨诸塞州波士顿计算机博物馆档案馆，“革命里程碑”文件。

65. Digital Equipment Corporation, “PDP-8 Typesetting System,” advertising brochure, 1966, Computer Museum Archives, Boston, MA, “Milestones of a Revolution” file.

66. Bell 等，《计算机工程》，第 215 页；Wes Clark，“LINC 成立较早，规模较小”，摘自 Adele Goldberg 主编的《个人工作站的历史：345–391》。

66. Bell et al., Computer Engineering, p. 215; Wes Clark, “The LINC Was Early and Small,” in Adele Goldberg, ed., A History of Personal Workstations: 345–391.

67.Bell等，64，180，198–199。

67. Bell et al., 64, 180, 198–199.

68.同上，第 198 页；另请参阅 Rifkin 和 Harrar 所著的《终极企业家》。

68. Ibid., 198; also Rifkin and Harrar, The Ultimate Entrepreneur.

69. Teletype Corporation，“Teletype，用于快速、经济的 8 级数据通信的 33 型设备”，产品文献，约 1966 年，AT&T 档案馆。

69. Teletype Corporation, “Teletype, Model 33 Equipment for Fast, Economical 8-level Data Communications,” Product Literature, ca. 1966, AT&T Archives.

70. CE MacKenzie，《编码字符集：历史与发展》（马萨诸塞州雷丁：Addison Wesley，1980 年）；Friden Corporation 的《2201 Flexowriter 自动书写机技术手册》，约 1968 年，副本归作者所有。

70. C. E. MacKenzie, Coded Character Sets: History & Development (Reading, MA: Addison Wesley, 1980); also Friden Corporation, “Technical Manual for 2201 Flexowriter Automatic Writing Machine,” ca. 1968, copy in author’s possession.

71.约翰·冷（John Leng），1993年4月13日与作者进行电话采访。

71. John Leng, telephone interview with the author, April 13, 1993.

72.描述类似关系的另一个常用术语是“增值经销商”（VAR）。

72. Another term that has come into common use, which describes a similar relationship, is “Value-Added Reseller” (VAR).

73.数字设备公司，《PDP-8 排版系统》，宣传册，约 1966 年；波士顿计算机博物馆，《革命里程碑》档案。

73. Digital Equipment Corporation, “PDP-8 Typesetting System,” brochure, ca. 1966; Computer Museum, Boston, “Milestones of a Revolution” Archive.

74. “歌舞线上：电脑灯光控制登陆百老汇”，戏剧工艺（1975 年 11 月至 12 月）：封面，第 6-11 页，第 26-29 页。一些剧院灯光专业人士私下告诉笔者，很多个晚上，当系统出现故障时，电脑会制造出一系列新的麻烦。

74. “A Chorus Line: Computerized Lighting Control Comes to Broadway,” Theatre Crafts (Nov.=Dec. 1975): cover, 6–11, 26–29. Some theater lighting professionals have privately told this author that the computer created a new set of headaches on many a night, when the system acted up.

75.波士顿计算机博物馆，“计算机历史时间轴”展览，1984-1989 年。

75. Computer Museum, Boston, “Timeline of Computer History” exhibition, 1984–1989.

76. Bob Cumberford，《对 Alec Issigonis 的采访》，《Road & Track 》（道路测试年鉴，1962 年）：18-19。

76. Bob Cumberford, “An Interview with Alec Issigonis,” Road & Track (Road Test Annual, 1962): 18–19.

77. DEC 的管理风格已成神话，尤其是在罗莎贝斯·莫斯·坎特 (Rosabeth Moss Kanter) 的《变革大师》(The Change Masters ) (1983) 出版之后。DEC 在 1990 年代的困境显然使人们开始质疑这一神话。作者对 DEC 衰落原因的看法将在后面的章节中探讨。

77. DEC’s management style attained the status of myth, especially after the publication of The Change Masters by Rosabeth Moss Kanter (1983). DEC’s troubles in the 1990s obviously lead one to question this myth. The author’s view on the causes of DEC’s downfall will be examined in a later chapter.

78.培生，《数字化工作》。

78. Pearson, Digital at Work.

79. Robert Sobel，IBM：转型中的巨人(纽约，1981)：228。

79. Robert Sobel, IBM: Colossus in Transition (New York, 1981): 228.

80.特鲁特·艾利森 (Truett Allison) 写给雷切尔·海伦加 (Rachel Hellenga) 的信，1990 年 9 月 14 日。波士顿计算机博物馆，“革命里程碑”档案。

80. Truett Allison to Rachel Hellenga, letter September 14, 1990. Computer Museum, Boston, “Milestones of a Revolution” Archive.

81. Ben Ross Schneider，《计算机世界之旅》（马萨诸塞州雷丁：Addison-Wesley，1974 年）：73。

81. Ben Ross Schneider, Travels in Computerland (Reading, MA: Addison-Wesley, 1974): 73.

82.参见 Ted Nelson 的《Computer Lib》（Ted Nelson，1974 年）；讽刺的是，在第二版的前言中，Nelson 指出，随着 DEC 的规模不断扩大，“它变得越来越像 IBM”。

82. See Ted Nelson, Computer Lib (Ted Nelson, 1974); ironically, in his preface to the second edition, Nelson notestext1d that as DEC grew larger, “it was getting just like IBM.”

83.索贝尔，IBM。

83. Sobel, IBM.

84. Max Palevsky，《口述历史访谈》，1973 年 2 月 15 日，第 20 页；史密森尼计算机历史项目；史密森尼档案馆。

84. Max Palevsky, Oral History Interview, February 15, 1973, 20; Smithsonian Computer History Project; Smithsonian Archives.

85. Pearson，《数字化工作》，第 1 章。第 2 章是经过编辑的对 Forrester 和 MITRE 公司的 Robert Everett 的采访记录。Forrester 和 Everett 都没有成为 DEC 的员工，尽管他们曾经或现在都是 DEC 的董事会成员。

85. Pearson, Digital at Work, chapter 1. Chapter 2 is an edited transcript of an interview with Forrester and Robert Everett, of the MITRE Corporation. Neither Forrester nor Everett became DEC employees, although both were, or still are, on the DEC Board of Directors.

86.硅谷的创业公司一直以来都大量聘用麻省理工学院的毕业生，现在也是如此，notestext1d 应该如此。数字设备公司未能保持竞争力的原因将在后面的章节中讨论，但仅凭这一点不应该导致 128 号公路作为创新中心的衰落。硅谷公司一直在失败，但该地区仍在繁荣发展。也许是波士顿风险投资家未能认识到创新理念并冒险投资；也可能是麻省理工学院未能维持其在 20 世纪 40 年代和 50 年代发展起来的卓越成就。

86. The entrepreneurial companies of Silicon Valley were always, and remain, heavily staffed by MIT graduates, it should be notestext1d. The failure of Digital Equipment Corporation to remain competitive will be discussed in a later chapter, but that alone should not have caused the fall of Route 128 as a center of innovation. Silicon Valley firms fail all the time, yet the region continues to prosper. Perhaps it was a failure by the Boston venture capitalists to recognize and risk money on innovative ideas; or it may have been a failure of MIT to sustain the excellence it had developed during the 1940s and 1950s.

87.约翰·肯尼斯·加尔布雷斯，《新工业国家》（波士顿：霍顿·米夫林，1967 年）；威廉·H·怀特，《组织人》（纽约：西蒙与舒斯特出版社，1956 年）。

87. John Kenneth Galbraith, The New Industrial State (Boston: Houghton Mifflin, 1967); William H. Whyte, The Organization Man (New York: Simon & Schuster, 1956).

第五章

Chapter 5

1.罗伯特·索贝尔，IBM：转型中的巨人（纽约：Bantam，1981年）：159、188、274。

1. Robert Sobel, IBM: Colossus in Transition (New York: Bantam, 1981): 159, 188, 274.

2. Franklin Fisher、James W. McKie 和 Richard B. Mancke，《IBM 和美国数据处理行业》（纽约：Praeger，1983 年）：65。

2. Franklin Fisher, James W. McKie, and Richard B. Mancke, IBM and the U.S. Data Processing Industry (New York: Praeger, 1983): 65.

3.同上；另请参阅 Sobel 著的《IBM》和 Katherine Fishman 著的《计算机机构》（纽约：麦格劳-希尔，1981 年）：第 470 页。

3. Ibid.; also Sobel, IBM, and Katherine Fishman, The Computer Establishment (New York: McGraw-Hill, 1981): 470.

4. Watts S. Humphrey，“MOBIDIC 和 Fieldata”，计算史年鉴9：2（1987）：137-182。

4. Watts S. Humphrey, “MOBIDIC and Fieldata,” Annals of the History of Computing 9: 2 (1987): 137–182.

5.鲍勃·埃文斯，引自 TA Wise 的《IBM 的 5,000,000,000 美元赌注》，《财富》（1966 年 9 月）：118–123、224、226、228。

5. Bob Evans, quoted in T. A. Wise, “IBM’s $5,000,000,000 Gamble,” Fortune (September 1966): 118–123, 224, 226, 228.

6.同上，第44页。

6. Ibid., 44.

7.相比之下，1975 年，施乐公司退出了大型计算机业务，注销了对前科学数据系统公司近 10 亿美元的投资——大约相当于 IBM 在 System=360 上投入的资金。这是美国公司面临的最大注销之一。然而，施乐公司幸存了下来。施乐进军计算机领域将在稍后讨论。

7. For comparison, in 1975, Xerox dropped out of the large computer business, writing off an investment in the former Scientific Data Systems of close to $1 billion—about what IBM had risked on System=360. If was one of the biggest write-offs ever faced by an American corporation. Yet Xerox survived. Xerox’s foray into computing will be discussed shortly.

8. Emerson Pugh、Lyle R. Johnson 和 John H. Palmer，《IBM 的 360 和早期 370 系统》（剑桥：麻省理工学院出版社，1991 年）：169。

8. Emerson Pugh, Lyle R. Johnson, and John H. Palmer, IBM’s 360 and Early 370 Systems (Cambridge: MIT Press, 1991): 169.

9.同上，第 169–174 页。另请参阅 Bob O. Evans 的《SPREAD 报告简介》，《计算机历史年鉴》第 5 期（1983 年）：第 4–5 页；Oscar H. Rothenbuecher 的《数据处理行业 50 强公司》，《Datamation 》（1976 年 6 月）：第 48–59 页。

9. Ibid., 169–174. See also Bob O. Evans, “Introduction to SPREAD Report,” Annals of the History of Computing 5 (1983): 4–5; also Oscar H. Rothenbuecher, “The Top 50 Companies in the Data Processing Industry,” Datamation (June 1976): 48–59.

10. Bob O. Evans，“SPREAD 报告：IBM System=360 项目的起源”，计算机史年鉴5（1983 年）：4-44。

10. Bob O. Evans, “SPREAD Report: the Origin of the IBM System=360 Project,” Annals of the History of Computing 5 (1983): 4–44.

11. Bob O. Evans，《IBM System=360》，《计算机博物馆报告》（1984 年夏季）：第 11 页。

11. Bob O. Evans, “IBM System=360,” Computer Museum Report (Summer 1984): 11.

12.同上，第11页。

12. Ibid., 11.

13. Minghui Hu，“维护 NASTRAN：航空航天计算的政治和技术”（硕士论文，弗吉尼亚州布莱克斯堡：弗吉尼亚理工大学，1995 年 2 月）。

13. Minghui Hu, “Maintaining NASTRAN: The Politics and Techniques of Aerospace Computing” (Masters Thesis, Blacksburg, VA: Virginia Polytechnic and State University, February 1995).

14. SPREAD 委员会报告由 13 人签署：John W. Haanstra、Bob O. Evans、Joel D. Aron、Frederick P. Brooks Jr.、John W. Fairclough、William P. Heisling、Herbert Hellerman、Walter H. Johnson、Michael J. Kelly、Douglas V. Newton、Bruce G. Oldfield、Deymour A. Rosen 和 Jerrold Svigals。

14. The SPREAD Committee report was signed by 13 men: John W. Haanstra, Bob O. Evans, Joel D. Aron, Frederick P. Brooks Jr., John W. Fairclough, William P. Heisling, Herbert Hellerman, Walter H. Johnson, Michael J. Kelly, Douglas V. Newton, Bruce G. Oldfield, Deymour A. Rosen, and Jerrold Svigals.

15. Evans，《传播报告》，第31页。

15. Evans, “Spread Report,” 31.

16. A. Padegs，“System=360 and Beyond”，IBM 研究与开发杂志25（1981）：377-390。

16. A. Padegs, “System=360 and Beyond,” IBM Journal of Research and Development 25 (1981): 377–390.

17. Maurice Wilkes，《设计自动计算机的最佳方法》，《曼彻斯特成立大会》，英国曼彻斯特（1951 年 7 月）：第 16-18 页。《计算史年鉴》第 8 卷重印（1986 年 4 月）：第 118-121 页。

17. Maurice Wilkes, “The Best Way to Design an Automatic Calculating Machine,” Manchester Inaugural Conference, Manchester, England (July 1951): 16–18. Reprinted in Annals of the History of Computing 8 (April 1986): 118–121.

18.这确实增加了一层复杂性，而且不可能不劳而获。SPREAD 委员会规定，如果 360 的设计者能够证明这样做至少比微编程设计的成本性能优势高出 1.25 倍，那么 360 的某个型号可以采用“硬连线”而不是微编程。与大多数超级计算机一样，360 的最高性能型号没有微编程。

18. This does add a layer of complexity, and one cannot get something for nothing. The SPREAD Committee specified that a model of the 360 could be “hard-wired” instead of microprogrammed if its designers could demonstrate that by doing so one gained at least a 1.25 cost-performance advantage over the microprogrammed design. The highest performance models of the 360, like most supercomputers, were not microprogrammed.

19.威尔克斯，《最佳方式》。

19. Wilkes, “The Best Way.”

20. Evans，《SPREAD 报告》，第33页。

20. Evans, “SPREAD Report,” 33.

21. Maurice Wilkes，《自动数字计算机》（伦敦：Methuen，1956 年）：139–140。

21. Maurice Wilkes, Automatic Digital Computers (London: Methuen, 1956): 139– 140.

22. PDP-8 小型计算机的设计者 Gordon Bell 称 360 通过改变微程序来模拟其他计算机的能力“可能是最重大的真正创新”：C. Gordon Bell 和 Allen Newell，《计算机结构》（纽约：McGraw-Hill，1971 年）：562。

22. Gordon Bell, the designer of the PDP-8 minicomputer, called the 360’s ability to emulate other computers through changing its microprogram “probably the most significant real innovation”: C. Gordon Bell and Allen Newell, Computer Structures (New York: McGraw-Hill, 1971): 562.

23. Pugh 等人，IBM 360，163，214–217。

23. Pugh et al., IBM’s 360, 163, 214–217.

24.此短语由 Alan Perlis 所言。

24. The phrase is attributed to Alan Perlis.

25. Pugh 等人，IBM 360，162–163。

25. Pugh et al., IBM’s 360, 162–163.

26. Buchholz，Werner 编辑；《规划计算机系统：延伸计划》（纽约，1962 年）；另请参阅 Fisher 等人的《IBM 和美国数据处理行业》，第 47-50 页。

26. Buchholz, Werner, ed; Planning a Computer System: Project Stretch (New York, 1962); see also Fisher et al., IBM and the U.S. Data Processing Industry, 47–50.

27. Buchholz，《Project Stretch》，第 40 页；另请参阅 Buchholz 的《字节词的起源》，《计算机史年鉴》第 3 卷（1981 年）：第 72 页。

27. Buchholz, Project Stretch, 40; also Buchholz, “Origin of the Word Byte,” Annals of the History of Computing 3 (1981): 72.

28. CEMacKenzie，《编码字符集：历史与发展》（马萨诸塞州雷丁：Addison Wesley，1980 年）：126 页。Mackenzie 是 IBM 的一名员工，也是 ASCII 开发委员会的成员。

28. C. E.MacKenzie, Coded Character Sets: History & Development (Reading, MA: Addison Wesley, 1980): 126 ff. Mackenzie was an employee of IBM who was a member of the committee developing ASCII.

29. Padegs，《IBM 研究与开发杂志》（1981 年 9 月）：379。

29. Padegs, IBM Journal of Research and Development (September 1981): 379.

30.比较一下 1984 年以后的两种标准个人电脑操作系统 MS-DOS 和 Macintosh 系统；以及 VHS 和 Beta，它们一直在争夺家用录像带的标准，直到前者占了上风。参见 Paul David，《理解 QWERTY 的经济学：历史的必要性》，载于《经济史与现代经济学家》（牛津，1986 年）：第 30-49 页。

30. Compare the two standard personal computer operating systems, MS-DOS and the Macintosh System, after 1984; with VHS and Beta, who battled for the standards in home videotapes, until the former prevailed. See Paul David, “Understanding the Economics of QWERTY: the Necessity of History,” in Economic History and the Modern Economist (Oxford, 1986): 30–49.

31.虽然没有直接关系，但还有另一个区别。有些计算机从内存中读取字节时，首先读取最高有效位字节；而有些计算机则相反。一组计算机称为“大端字节序”，另一组计算机称为“小端字节序”。这些术语来自《格列佛游记》，其中描述了一场争夺煮鸡蛋哪一面的冲突。参见 Danny Cohen 的《论圣战与和平恳求》，IEEE 计算机14（1981 年 10 月）：48-54。

31. Although not directly related, there is another difference. Some computers fetch bytes from memory with the most significant first; others do the reverse. One group is called “big-endian,” the other “little-endian.” The terms are from Gulliver’s Travels, which describes a conflict over which side of a boiled egg to open. See Danny Cohen, “On Holy Wars and a Plea for Peace,” IEEE Computer 14 (October 1981): 48–54.

32.这种现象部分源于设计师的短视。事实证明，360 不仅比预期更受欢迎，而且其架构也经久不衰。因此，对更多地址位的需求是衡量架构成功的标准。数字设备公司的 PDP-11 小型计算机和英特尔公司的 8086 系列微处理器也发生了同样的情况。参见 NS Prasad，《IBM 大型机：架构与设计》（纽约：麦格劳希尔，1989 年）。

32. This phenomenon stems only in part from a short-sightedness on the designers’ part. The 360 proved not only to be more popular than expected; its architecture proved much more long-lasting as well. The need for more address bits is thus a measure of an architecture’s success. The same course of events happened with the Digital Equipment Corporation’s PDP-11 minicomputer, and with the Intel Corporations’s 8086 series of microprocessors. See N. S. Prasad, IBM Mainframes: Architecture and Design (New York: McGraw Hill, 1989).

33. Evans，“SPREAD 报告”，第 31 页；另请参阅 Evans，“IBM System=360”。

33. Evans, “SPREAD Report,” 31; also Evans, “IBM System=360.”

34.最初发布时还包含一个 I=O 设备，即视频显示终端。2250 型并不是一个成功的产品，但后来的 VDT 3270 型却很成功。视频终端的使用并不像 360 型的打卡机和读卡器那样频繁。

34. One I=O device also included in the initial announcement was a video display terminal. The Model 2250 was not a successful product, but a later VDT, the Model 3270, was. Video terminals were not used as much as card punches and readers for the 360.

35. Fernando Corbato，“实验性分时系统”，Proc. SJCC，旧金山，1962 年 5 月 1-3 日（AFIPS Press，1962 年）：335-344。

35. Fernando Corbato, “An Experimental Time-sharing System,” Proc. SJCC, San Francisco, May 1–3, 1962 (AFIPS Press, 1962): 335–344.

36. Judy E. O'Neill，“通过分时和网络实现交互式计算的演进”(博士论文，明尼苏达大学，1992年)。

36. Judy E. O’Neill, “The evolution of interactive computing through timesharing and networking” (Ph.D. diss., University of Minnesota, 1992).

37. Pugh 等人，IBM 的 360 和早期的 370 系统，360–363。

37. Pugh et al., IBM’s 360 and Early 370 Systems, 360–363.

38.同上，第360–363页。

38. Ibid., 360–363.

39.在 SPREAD 报告的口述历史中（Evans，《SPREAD 报告》，1983 年），采访者和 SPREAD 委员会前成员就这一点进行了激烈的交流。我的论点部分来自这次讨论，但也来自后面提供的其他证据。

39. In the oral histories of the SPREAD Report (Evans, “The SPREAD Report,” 1983), there is a vigorous exchange between the interviewers and the former members of the SPREAD Committee over this point. My arguments follow in part from this discussion but also from other evidence, as given later.

40. Arthur Norberg 和 Judy O'Neill，《计算机技术的变革：五角大楼的信息处理，1962-1986》（巴尔的摩：约翰霍普金斯大学出版社，1996 年）：第 100 页。Norberg 和 O'Neill 没有详细阐述这一标准；在上述《计算机史年鉴》的采访对象中，有几位人士强烈反对麻省理工学院对 System=360 所谓缺陷的评估。

40. Arthur Norberg and Judy O’Neill, Transforming Computer Technology: Information Processing for the Pentagon, 1962–1986 (Baltimore: Johns Hopkins University Press, 1996): 100. Norberg and O’Neill do not elaborate on this criterion; among those interviewed for the Annals of the History of Computing, cited above, there were several who vigorously disagreed with MIT’s assessment of the System=360’s alleged deficiencies.

41. Melinda Varian，《VM 和 VM 社区：过去、现在和未来》，未发表的打字稿（新泽西州普林斯顿，1991 年 4 月）：23-24。

41. Melinda Varian, “VM and the VM Community: Past, Present, and Future,” unpublished typescript (Princeton, NJ, April 1991): 23–24.

42. Sobel，IBM：《转型中的巨人》，284–286。

42. Sobel, IBM: Colossus in Transition, 284–286.

43. Pugh 等人，IBM 的 360 和早期的370，448–450。

43. Pugh et al., IBM’s 360 and Early 370, 448–450.

44.同上，第451页。

44. Ibid., 451.

45.后缀“tron”源于希腊语“犁”，通常指仪器或探测器。

45. The suffix “tron” is descended from the Greek word “to plow,” and typically referred to an instrument or probe.

46.约翰·布鲁克斯，《疯狂年代》（纽约：达顿出版社，1984 年）。他还给出了另一种定义：“一种操作股票市场的方法……特点是快速买卖大量股票。”

46. John Brooks, The Go-Go Years (New York: Dutton, 1984). He also gives an alternate definition: “a method of operating the stock market ... characterized by rapid in-and-out trading of huge blocks of stock.”

47. Brooks，Go-Go，231；Fishman，Computer Establishment，246-249；Sobel，IBM，224-225。

47. Brooks, Go-Go, 231; Fishman, Computer Establishment, 246–249; Sobel, IBM, 224–225.

48. Evans，《IBM 系统 = 360》，《计算机博物馆报告》（1984 年夏季）：第 18 页。熊市总体以及联邦反垄断诉讼也是因素之一。

48. Evans, “IBM System=360,” Computer Museum Report (Summer 1984): 18. The general bear market, as well as the Federal antitrust suit, were also factors.

49. Brooks，引自Evans，《SPREAD 报告》，第 31 页。

49. Brooks, quoted in Evans, “The SPREAD Report,” 31.

50. Boelie Elzen 和 Donald MacKenzie，“速度的社会限制”，计算史年鉴16：1（1994）：46-61。

50. Boelie Elzen and Donald MacKenzie, “The Social Limits of Speed,” Annals of the History of Computing 16: 1 (1994): 46–61.

51.同上，另见 Seymour Cray 的“砷化镓到底是怎么回事？”演讲录像带，佛罗里达州奥兰多，1988 年 11 月。

51. Ibid., also Seymour Cray, “What’s all this about Gallium Arsenide?” videotape of a talk, Orlando, FL, November 1988.

52. Saul Rosen，《电子计算机：历史综述》，《计算综述》 1（1969 年 3 月）：21。

52. Saul Rosen, “Electronic Computers: a Historical Survey,” Computing Surveys 1 (March 1969): 21.

53. Fishman，《计算机机构》，182。

53. Fishman, The Computer Establishment, 182.

54.Sobel，IBM，288；Fishman，计算机机构，165。

54. Sobel, IBM, 288; Fishman, Computer Establishment, 165.

55. Jamie Pearson，《数字化工作》（马萨诸塞州贝德福德：Digital Press，1992 年）：127–128。

55. Jamie Pearson, Digital at Work (Bedford, MA: Digital Press, 1992): 127–128.

56. Pugh 等人，IBM 的 360 和早期的 370 系统，551。

56. Pugh et al., IBM’s 360 and Early 370 Systems, 551.

57. Fisher 等人，《IBM 和美国数据处理行业》，379。

57. Fisher et al., IBM and the U.S. Data Processing Industry, 379.

58. Kenneth Flamm，创造计算机：政府、工业和高科技(华盛顿特区：布鲁金斯学会，1988 年）：132、195-196。

58. Kenneth Flamm, Creating the Computer: Government, Industry, and High Technology (Washington, DC: Brookings Institution, 1988): 132, 195–196.

59.有关这个时代与富兰克林·费舍尔截然不同的观点，请参阅 Thomas DeLamarter 的《蓝色巨人：IBM 的权力运用与滥用》（纽约：Dodd, Mead，1986 年）。

59. For a view of this era quite different from Franklin Fisher’s, see Thomas DeLamarter, Big Blue: IBM’s Use and Abuse of Power (New York: Dodd, Mead, 1986).

60. “1106”、“1108”等命名法让人想起 UNIVAC 作为 20 世纪 50 年代第一批科学计算机之一 1103 的供应商所取得的成就。

60. The nomenclature “1106,” “1108,” and so on was a reminder of UNIVAC’s legacy as the supplier of one of the first scientific computers in the 1950s: the 1103.

61.斯佩里兰德公司，《斯佩里事实》，1979 年宣传册。

61. Sperry Rand Corporation, “Sperry Facts,” brochure dated 1979.

62.马克斯·帕列夫斯基，1973 年 2 月 15 日接受 R. Mapstone 的采访，史密森尼计算机历史项目，史密森尼档案馆。

62. Max Palevsky, interview with R. Mapstone, February 15, 1973, Smithsonian Computer History Project, Smithsonian Archives.

63.同上，第 12–13 页。PB-250 的售价为 25,000 至 30,000 美元。其最独特的架构特征是使用延迟线存储器 — 可能是唯一使用这种存储器的晶体管机器。

63. Ibid., 12–13. The PB-250 sold for $25,000 to $30,000. Its most distinctive architectural feature was its use of a delay line memory—possibly the only transistorized machine to use one.

64.同上，另请参阅 Fishman 的《计算机机构》，第 216-219 页。

64. Ibid., also Fishman, The Computer Establishment, 216–219.

65.道格拉斯·史密斯，《摸索未来：施乐公司如何发明却又忽视了第一台个人电脑》（纽约：William Morrow，1988 年）：122。

65. Douglas Smith, Fumbling the Future: How Xerox Invented, Then Ignored, the First Personal Computer (New York: William Morrow, 1988): 122.

66. Fisher 等人，《IBM 和美国数据处理行业》，267。

66. Fisher et al., IBM and the U.S. Data Processing Industry, 267.

67.史密森学会，计算机历史项目，马克斯·帕列夫斯基论文，史密森学会档案馆。SDS-940 分时系统基于价值 25 万美元的 SDS-930，该公司为其添加了专用内存、输入输出单元和软件。

67. Smithsonian Institution, Computer History Project, Max Palevsky Papers, Smithsonian Archives. The SDS-940 time-sharing system was based on the quarter-million-dollar SDS-930, to which the company added specialized memory, Input=Output units, and software.

68. Hoelzer，1993 年 6 月 25 日与作者的电话采访；Charles Bradshaw，1993 年 7 月 21 日写给作者的信；另见 LA Lanzarotta，《亨茨维尔的计算能力》，Datamation（1960 年 7 月至 8 月）：18-21。

68. Hoelzer, telephone interview with author, 25 June 1993; Charles Bradshaw, letter to the author 21 July 1993; also L. A. Lanzarotta, “Computing Power at Huntsville,” Datamation (July=August 1960): 18–21.

69. Smith，《摸索未来》；另请参阅 Fishman，《计算机机构》，222；另请参阅 Max Palevsky 的证词，1974 年 7 月 10 日；哈格利博物馆，编号 #1912，盒子 31，文件夹 0990–0993。

69. Smith, Fumbling the Future; also Fishman, Computer Establishment, 222; also Max Palevsky deposition, 10 July 1974; Hagley Museum, Accession #1912, box 31, folder 0990–0993.

70.服务局是一家公司或公司的一个部门，它通过软件向客户提供帮助或者通过其他方式帮助客户获得计算机来解决客户的特定问题。

70. A service bureau is a company or division of a company that assists a customer with software and otherwise helps to get a computer to solve a customer’s specific problems.

71. Elmer Kubie，《对第一家软件公司的反思》，《计算机史年鉴》，16：2（1994）：65-71；另见 Walter F. Bauer 1983 年 5 月 16 日与 Arthur Norberg 的访谈，查尔斯·巴贝奇研究所档案。

71. Elmer Kubie, “Reflections of the First Software Company,” Annals of the History of Computing, 16: 2 (1994): 65–71; also Walter F. Bauer, interview with Arthur Norberg, 16 May 1983, Charles Babbage Institute Archives.

72. “自动数据处理公司，”胡佛手册（1991）：EMM-102。

72. “Automatic Data Processing, Inc.,” Hoover’s Handbook (1991): EMM-102.

73. “ADP”，国际公司历史目录（1988）：117-119。

73. “ADP,” International Directory of Company Histories (1988): 117–119.

74. Fishman，《计算机机构》，273。

74. Fishman, The Computer Establishment, 273.

75. Franklin Fisher、John J. McGowan 和 Joen E. Greenwood，《折叠、纺锤和残缺：经济分析和美国与 IBM 案》（剑桥：麻省理工学院出版社，1983 年）：204-218。

75. Franklin Fisher, John J. McGowan, and Joen E. Greenwood, Folded, Spindled, and Mutilated: Economic Analysis and U.S. vs. IBM (Cambridge: MIT Press, 1983): 204–218.

76. Fred Brooks，《人月神话：软件工程随笔》（马萨诸塞州雷丁：Addison-Wesley，1975 年）。

76. Fred Brooks, The Mythical Man-Month: Essays on Software Engineering (Reading, MA: Addison-Wesley, 1975).

77. Fisher 等人，IBM 和美国数据处理行业，176-178。

77. Fisher et al., IBM and the U.S. Data Processing Industry, 176–178.

78. Robert Slater，《硅中的肖像》（剑桥：麻省理工学院出版社，1992 年）；第 12 章。

78. Robert Slater, Portraits in Silicon (Cambridge: MIT Press, 1992); chapter 12.

79. Claude Baum，《系统构建者：SDC 的故事》（加利福尼亚州圣莫尼卡：系统开发公司，1981 年）。

79. Claude Baum, The System Builders: the Story of SDC (Santa Monica, CA: System Development Corporation, 1981).

80. TRW 年度报告，NASM 技术文件。

80. Annual Reports of TRW, NASM Tech Files.

81. 《华盛顿邮报》，1994年2月9日，c-1。

81. The Washington Post, February 9, 1994, c-1.

82. Fisher，McGowan 和 Greenwood，《Folded, Spindled, and Mutilated》，第 1 章。

82. Fisher, McGowan, and Greenwood, Folded, Spindled, and Mutilated, chapter 1.

83. Datamation，行业调查，1986 年 6 月 16 日；1987 年 6 月 15 日。

83. Datamation, Industry surveys, June 16, 1986; June 15, 1987.

84. Kenneth Flamm，《针对计算机：政府支持和国际竞争》（华盛顿特区：布鲁金斯学会，1987 年）：77-78。

84. Kenneth Flamm, Targeting the Computer: Government Support and International Competition (Washington, DC: Brookings Institution, 1987): 77–78.

85. Lundstrom，《来自 UNIVAC 的几个好人》，225。部分商业信贷后来被出售给邓白氏。

85. Lundstrom, A Few Good Men From UNIVAC, 225. Portions of Commercial Credit were later sold to Dun & Bradstreet.

86. James Worthy，“控制数据公司：诺里斯时代”，计算机史年鉴17:1（1995）：47-53。

86. James Worthy, “Control Data Corporation: the Norris Era,” Annals of the History of Computing 17: 1 (1995): 47–53.

87. Lundstrom，《来自 UNIVAC 的几个好人》。

87. Lundstrom, A Few Good Men From UNIVAC.

88.关于柏拉图的最佳描述可以在特德·纳尔逊的《计算机解放》（特德·纳尔逊，1974）：DM26-27中找到。

88. The best description of PLATO is found in Ted Nelson, Computer Lib (Ted Nelson, 1974): DM26–27.

第六章

Chapter 6

1. Herb Grosch，《计算机：生活中的点滴片段》（加利福尼亚州诺瓦托：Third Millenium Books，1991 年），第 13 章。该定律的严格表述是“计算能力随着成本的平方而增加”。

1. Herb Grosch, Computer: Bit Slices from a Life (Novato, CA: Third Millenium Books, 1991), chapter 13. The strict statement of the law is that “Computing power increases as the square of the cost.”

2.有关该定律是否正确的一些争论，请参阅 Herb Grosch 的《Grosch 定律再探》，《Computerworld》（1975 年 4 月 16 日）：24；另请参阅 Phillip Ein-Dor 的《Grosch 定律再探》，《CACM》 28：2（1985 年 2 月）：142-151。

2. For some of the flavor of the debate over whether this law was in fact true, see Herb Grosch, “Grosch’s Law Revisited,” Computerworld (April 16, 1975): 24; also Phillip Ein-Dor, “Grosch’s Law Re-Revisited,” CACM 28: 2 (February 1985): 142– 151.

3.当时在普林斯顿建造的高等研究院计算机是一个很好的例子，说明如果不使用这种方法，事情会变得多么困难。由于它基于冯·诺依曼原理，因此所需的电子管比 ENIAC 要少得多。但是当一个电子管烧坏时，工程师通常需要剪断一团乱麻般的电线才能找到电子管，拔掉插头，再换上新的。无论设计多么优雅，这都不是一个优雅的实现。IAS 计算机在史密森学会展出。

3. The Institute for Advanced Study Computer, built in Princeton at that time, is a good example of how difficult things are when this approach is not used. Because it was based on von Neumann’s principles, it required far fewer tubes than the ENIAC. But when a tube burned out, it was often necessary for an engineer to cut through a spaghetti tangle of wires simply to get at the tube, unplug it, and put in a fresh one. Hardly an elegant implementation, however elegant the design. The IAS computer is on exhibit at the Smithsonian Institution.

4. Charles J. Bashe、Lyle R. Johnson、John H. Palmer 和 Emerson W. Pugh，《IBM 的早期计算机》（剑桥：麻省理工学院出版社，1986 年）：408-413。

4. Charles J. Bashe, Lyle R. Johnson, John H. Palmer, and Emerson W. Pugh, IBM’s early Computers (Cambridge: MIT Press, 1986): 408–413.

5. Ernest Braun 和 Stuart Macdonald，《微型革命：半导体电子学的历史和影响》，第 2 版（英国剑桥：剑桥大学出版社，1982 年）；Michael F. Wolff，《集成电路的起源》，《IEEE 频谱》（1976 年 8 月）：第 45-53 页。

5. Ernest Braun and Stuart Macdonald, Revolution in Miniature: the History and Impact of Semiconductor Electronics, 2nd ed. (Cambridge, UK: Cambridge University Press, 1982); also Michael F. Wolff, “The Genesis of the IC,” IEEE Spectrum (August 1976): 45–53.

6. Braun 和 Macdonald，《微型革命》，各处均有提及；此外，Herbert Kleiman 还著有《集成电路：电子行业产品创新的案例研究》（博士论文，乔治华盛顿大学，1966 年）。

6. Braun and Macdonald, Revolution in Miniature, passim; also Herbert Kleiman, “The Integrated Circuit: a Case Study in Product Innovation in the Electronics Industry” (Ph.D. diss., George Washington University, 1966).

7. Bell 被引用于IEEE Spectrum 25（1988 年 11 月）：87。

7. Bell is quoted in IEEE Spectrum 25 (November 1988): 87.

8. Robert Noyce，《微电子学》，《科学美国人》（1977年9月）：64。

8. Robert Noyce, “Microelectronics,” Scientific American (September 1977): 64.

9. Cledo Brunetti 和 Roger W. Curtis，“印刷电路技术”，通函#468（华盛顿特区：国家标准局，1947 年）。

9. Cledo Brunetti and Roger W. Curtis, “Printed Circuit Techniques,” Circular #468 (Washington, DC: National Bureau of Standards, 1947).

10. Paul Eisler，《我与印刷电路的生活》，由 Mari Williams 编辑（宾夕法尼亚州伯利恒：利哈伊大学出版社，1989 年出版）；但请参阅 Charles Susskind 在《Science》第 247 期（1990 年 2 月 23 日）：第 986 页对 Eisler 的评论。

10. Paul Eisler, My Life with the Printed Circuit, edited with notestext1s by Mari Williams (Bethlehem, PA: Lehigh University Press, 1989); but see the review of Eisler by Charles Susskind in Science 247 (23 February 1990): 986.

11. Eisler，专利号#2,441,960，1948 年 5 月 25 日；申请日期为 1944 年 2 月 3 日。

11. Eisler, patent #2,441,960, May 25, 1948; applied for February 3, 1944.

12. Eisler，《我的一生》；另请参阅 Thomas Misa，《军事需求、商业现实和晶体管的发展，1948-1958》，收录于 Merritt Roe Smith 主编的《军事企业和技术变革》（剑桥：麻省理工学院出版社，1985 年）：第 253-287 页。

12. Eisler, My Life; also Thomas Misa, “Military Needs, Commercial Realities, and the Development of the Transistor, 1948–1958,” in Merritt Roe Smith, ed., Military Enterprise and Technological Change (Cambridge: MIT Press, 1985): 253– 287.

13. Jack S. Kilby，“集成电路的发明”，IEEE 电子器件学报23（1976 年 7 月）：648-654。

13. Jack S. Kilby, “Invention of the Integrated Circuit,” IEEE Transactions on Electron Devices 23 (July 1976): 648–654.

14. Braun 和 Macdonald，《微型革命》，第 95 页；另请参阅 Kleiman，《集成电路：案例研究》，第 111 页。

14. Braun and Macdonald, Revolution in Miniature, 95; also Kleiman, “The Integrated Circuit: a Case Study,” 111.

15. 航空周刊（1961 年 5 月 29 日）：82-83。

15. Aviation Week (May 29, 1961): 82–83.

16. Larry Waller，《洁净室发明家Whitfield留下了无可挑剔的遗产》，《电子学》（1985年2月4日）：38。

16. Larry Waller, “Clean-Room Inventor Whitfield Leaves a Spotless Legacy,” Electronics (February 4, 1985): 38.

17. “民兵高可靠性零部件项目：历史与遗产”，罗克韦尔国际公司，Autonetics战略系统分部，加利福尼亚州阿纳海姆，报告 C81-451=201，1981 年 7 月 31 日；NASM 档案。

17. “The Minuteman High Reliability Component Parts Program: a History and Legacy,” Rockwell International, Autonetics Strategic Systems Division, Anaheim, CA, Report C81-451=201, July 31, 1981; NASM Archives.

18. Philip J. Klass，《可靠性是民兵的基本目标》，《航空周刊》（1959 年 10 月 19 日）：第 13F 页。

18. Philip J. Klass, “Reliability is Essential Minuteman Goal,” Aviation Week (October 19, 1959): 13F.

19.詹姆斯·韦伯，引自克莱曼的《集成电路》，第 72 页；另见《新闻周刊》（1971 年 7 月 26 日）：第 13 页；以及《导弹与火箭》（1963 年 6 月 24 日）。将单个民兵导弹的单位成本缩小到该范围之外是不切实际的。

19. James Webb, quoted by Kleiman, “The Integrated Circuit,” 72; also Newsweek (July 26, 1971): 13; and Missiles and Rockets (June 24, 1963). It is not practical to narrow the per-unit cost of a single Minuteman beyond that range.

20.布劳恩和麦克唐纳引用了一位工程师的话，他说，如果所有军用电子元件都以民兵那样精心采购，那么它们的总成本将超过美国的国民生产总值。（《微型革命》，99）。

20. Braun and Macdonald quote an engineer who stated that if all military electronic components were acquired with the care lavished on Minuteman, their combined costs would exceed the U.S. GNP. (Revolution in Miniature, 99).

21.唐纳德·麦肯齐，《发明准确性：核导弹制导的历史社会学》（剑桥，麻省理工学院出版社，1990 年）：155。

21. Donald MacKenzie, Inventing Accuracy: A Historical Sociology of Nuclear Missile Guidance (Cambridge, MIT Press, 1990): 155.

22. Jack S. Kilby，《集成电路的发明》，IEEE 电子设备汇刊第 23 卷（1976 年 7 月）：648–654 页；另请参阅 Kilby 的《传记资料》，作者文集中的 MS 版。

22. Jack S. Kilby, “Invention of the Integrated Circuit,” IEEE Transactions on Electron Devices 23 (July 1976): 648–654; also Kilby, “Biographical Data,” MS in the author’s collection.

23. Kilby，《集成电路的发明》，第 650 页。

23. Kilby, “Invention of the Integrated Circuit,” 650.

24.Kilby，“传记数据”。

24. Kilby, “Biographical Data.”

25. TR Reid，《芯片》（纽约：西蒙与舒斯特出版社，1985 年）：57-58。

25. T. R. Reid, The Chip (New York: Simon & Schuster, 1985): 57–58.

26. RCA 是 Micro-Module 资金的主要接受者。参见 Braun 和 Macdonald 著《Revolution in Miniature》，第 95 页。

26. RCA was the primary recipient of funds for Micro-Module. See Braun and Macdonald, Revolution in Miniature, 95.

27.美国专利 3,138,743，“微型电子电路”。

27. U.S. Patent 3,138,743 for “Miniaturized Electronic Circuits.”

28. Kilby，“集成电路的发明”，650–651 页；另请参阅 Kilby，“传记数据”。

28. Kilby, “Invention of the I.C.,” 650–651; also Kilby, “Biographical Data.”

29. Tom Wolfe，《罗伯特·诺伊斯的修补》，《Esquire》（1983 年 12 月：346-374；另请参阅 Robert Knapp 和 HB Goodrich 合著的《美国科学家的起源》（芝加哥：芝加哥大学出版社，1952 年）。

29. Tom Wolfe, “The Tinkerings of Robert Noyce,” Esquire (December 1983: 346–374; also Robert Knapp and H. B. Goodrich, The Origins of American Scientists (Chicago: University of Chicago Press, 1952).

30. Michael Wolff，《集成电路的起源》，IEEE Spectrum（1976 年 8 月）：45-53。

30. Michael Wolff, “The Genesis of the IC,” IEEE Spectrum (August 1976): 45–53.

31. Carolyn Caddes，《成功肖像：硅谷先驱者的印象》（加利福尼亚州帕洛阿尔托：蒂奥加，1986 年）：第 44-45 页。

31. Carolyn Caddes, Portraits of Success: Impressions of Silicon Valley Pioneers (Palo Alto, CA: Tioga, 1986): 44–45.

32. Wolff，《集成电路的起源》，第 51 页。

32. Wolff, “The Genesis of the IC,” 51.

33.托马斯·J·沃森写给谢尔曼·费尔柴尔德的信，1926 年 2 月 27 日；谢尔曼·费尔柴尔德文件，国会图书馆，第 29 盒。

33. Letter from Thomas J. Watson to Sherman Fairchild, February 27, 1926; Sherman Fairchild Papers, Library of Congress, Box 29.

34.转引自Caddes著《成功的肖像》，第44页。

34. Quoted in Caddes, Portraits of Success, 44.

35. Eugene S. Ferguson，《工程与心灵之眼》（剑桥：麻省理工学院出版社，1992 年）；该书是 Ferguson 的文章“心灵之眼：技术中的非语言思维”的扩展版本，《科学》 197（1977 年 8 月 26 日）：827–836。

35. Eugene S. Ferguson, Engineering and the Mind’s Eye (Cambridge: MIT Press, 1992); this book is an expanded version of an article by Ferguson, “The Mind’s Eye: Nonverbal Thought in Technology,” Science 197 (26 August 1977): 827–836.

36.弗格森哀叹工程专业学生的这种技能正在逐渐丧失，部分原因是教授们希望让工程课程更加“科学化”。弗格森列举了最近工程失败的几个例子来证明自己的观点，但他没有认识到集成电路设计技术可能是一个例外。我对 Apple II 和 Data General Nova 等计算机设计的理解可能与弗格森的论点相矛盾，后面会讨论这些设计。

36. Ferguson laments what he sees as the loss of this skill among engineering students, in part fostered by professors who wish to make the engineering curriculum more “scientific.” Ferguson gives a number of examples of recent engineering failures to make his point, but he does not recognize the technology of integrated circuit design as a possible exception. My understanding of the design of computers like the Apple II and Data General Nova, discussed later, may contradict Ferguson’s argument.

37.Kleiman，“集成电路”。

37. Kleiman, “The Integrated Circuit.”

38. Marth Smith Parks，《20 世纪 70 年代的微电子学》（加利福尼亚州阿纳海姆，罗克韦尔国际公司，1974 年）：第 64 页。民兵 1 号导弹通过在其发射井中旋转整个导弹来重新瞄准目标；从中可以推断出那些目标可能是什么，以及它们改变的频率有多低。

38. Marth Smith Parks, “Microelectronics in the 1970’s” (Anaheim, CA, Rock-well International, 1974): 64. The Minuteman I was retargeted by physically rotating the entire missile in its silo; from that one can infer what those targets might have been, and how infrequently they were changed.

39. Jack Kilby，写给 Gwen Bell 的信，1984 年 6 月 26 日；波士顿；计算机博物馆档案馆。

39. Jack Kilby, letter to Gwen Bell, June 26, 1984; Boston; Computer Museum Archives.

40. 《导弹与火箭》（1964 年 3 月 2 日）：第 35 页；《航空周刊与空间技术》（1965 年 8 月 26 日）。

40. Missiles and Rockets (March 2, 1964): 35; Aviation Week and Space Technology (August 26, 1965).

41. “1964年：微电路成长的一年”，电子学第37卷（1964年3月13日）：第10-11页。

41. “Nineteen Sixty-Four: the Year Microcircuits Grew Up,” Electronics 37 (March 13, 1964): 10–11.

42.转引自 Dirk Hanson 著《新炼金术士：硅谷与微电子革命》（波士顿：Little, Brown，1982 年）：第 93 页。

42. Quoted in Dirk Hanson, The New Alchemists: Silicon Valley and the Microelectronics Revolution (Boston: Little, Brown, 1982): 93.

43. Robert Noyce，“军事装备中的集成电路”，IEEE Spectrum（1964 年 6 月）：71。

43. Robert Noyce, “Integrated Circuits in Military Equipment,” IEEE Spectrum (June 1964): 71.

44. Paul E. Ceruzzi，《超越极限：飞行进入计算机时代》（剑桥：麻省理工学院出版社，1989 年），第 6 章。

44. Paul E. Ceruzzi, Beyond the Limits: Flight Enters the Computer Age (Cambridge: MIT Press, 1989), chapter 6.

45. Donald C. Fraser 和 Philip Felleman，“数字电传操纵系统：计算机引领潮流”，宇航与航空学报12：7=8（1974）：24-32。

45. Donald C. Fraser, and Philip Felleman, “Digital Fly-by-Wire: Computers Lead the Way,” Astronautics and Aeronautics 12: 7=8 (1974): 24–32.

46. Eldon C. Hall，《月球之旅：阿波罗制导计算机的历史》（弗吉尼亚州雷斯顿：AIAA，1996 年）：第 82 页；另见 A. Michal McMahon 的《计算机与综合体：战后美国技术创新研究》，1986 年 10 月，美国宇航局历史办公室，华盛顿特区，第 30 页。

46. Eldon C. Hall, Journey to the Moon: the History of the Apollo Guidance Computer (Reston, VA: AIAA, 1996): 82; also A. Michal McMahon, “The Computer and the Complex: a Study of Technical Innovation in Postwar America,” October 1986, NASA History Office, Washington, DC, 30.

47. Eldon Hall，《阿波罗制导计算机：设计师的观点》，波士顿计算机博物馆报告（1982 年秋季）：2-5。

47. Eldon Hall, “The Apollo Guidance Computer: a Designer’s View,” Computer Museum, Boston, Report (Fall 1982): 2–5.

48.具有三个输入的 NOR 门。该芯片包含三个晶体管和四个电阻器。

48. A NOR-gate with three inputs. The chip contained three transistors and four resistors.

49. James Tomayko，《太空飞行中的计算机：NASA 的经验》 (华盛顿特区：NASA 承包商报告 182505，1988 年)：第 28-30 页。

49. James Tomayko, “Computers in Spaceflight: the NASA Experience,” (Washington, DC: NASA Contractor Report 182505, 1988): 28–30.

50. A. Michal McMahon，“计算机与综合体：战后美国技术创新研究”

50. A. Michal McMahon, “The Computer and the Complex: a Study of Technical Innovation in Postwar America.”

51. Pugh 等人，IBM 的 360 和早期 370 系统，76-83；EM Davis 等人，《固态逻辑技术：多功能高性能微电子技术》，IBM 杂志，8（1964 年 4 月）：102-114。

51. Pugh et al., IBM’s 360 and Early 370 Systems, 76–83; also E. M. Davis et al., “Solid Logic Technology: Versatile, High Performance Microelectronics,” IBM Journal, 8 (April 1964): 102–114.

52.第一个引文来自 Bob Henle，引自 Pugh 等人，第 105 页。第二个引文来自 John Haanstra，《Monolithics and IBM》，1964 年 9 月报告，未分页，IBM 档案馆，纽约瓦尔哈拉。感谢 Emerson Pugh 为我提供此文档的副本。

52. The first quotation is from Bob Henle, quoted in Pugh et al., 105. The second is from John Haanstra, “Monolithics and IBM,” report of September 1964, unpaginated, IBM Archives, Valhalla, NY. I am grateful to Emerson Pugh for providing me with a copy of this document.

53. Pugh 等，IBM 的 360 和早期的 370；C. Gordon Bell，《小型和微型工业》，IEEE计算机（1984 年 10 月）：14-29；Datamation（1968 年 11 月）：72-73；Datamation（1974 年 7 月）：50-60。

53. Pugh et al., IBM’s 360 and Early 370; C. Gordon Bell, “The Mini and Micro Industries,” IEEE Computer (October 1984): 14–29; Datamation (November 1968): 72–73; Datamation (July 1974): 50–60.

54. Bell，“小型和微型产业”，14-29。

54. Bell, “The Mini and Micro Industries,” 14–29.

55.唐·兰开斯特（Don Lancaster），TTL 食谱（印第安纳波利斯：Howard Sams，1974 年）。

55. Don Lancaster, TTL Cookbook (Indianapolis: Howard Sams, 1974).

56. IEEE Spectrum 25: 11 (1970): 70；另请参阅 Tom Monte 和 Ilene Pritikin 的《Pritikin：治愈美国心灵的人》（宾夕法尼亚州 Emmaus: Rodale Press, 1988）。

56. IEEE Spectrum 25: 11 (1970): 70; also Tom Monte and Ilene Pritikin, Pritikin: the Man who Healed America’s Heart (Emmaus, PA: Rodale Press, 1988).

57. “SYMBOL：一个探索主要硬件替代软件的大型实验系统”，Daniel Siewiorek、C. Gordon Bell 和 Allen Newell 编，《计算机结构：原理与范例》（纽约：McGraw-Hill，1982 年）：489-507。

57. “SYMBOL: A Large Experimental System Exploring Major Hardware Replacement of Software,” in Daniel Siewiorek, C. Gordon Bell, and Allen Newell, eds. Computer Structures: Principles and Examples (New York: McGraw-Hill, 1982): 489– 507.

58.主要变化是“互补金属氧化物半导体”（CMOS）取代了 TTL，DIP 外壳逐渐被“单列直插式内存模块”（SIMM）取代，以及笔记本电脑等产品的扁平包装。

58. Major changes have been the advent of “Complementary Metal-Oxide Semiconductor” (CMOS) in place of TTL, and the gradual replacement of the DIP housing to “Single In-Line Memory Modules” (SIMM), and flat packaging for products like laptops.

59. W. Buchholz，《字节词的起源》，计算机史年鉴10：4（1989）：340。

59. W. Buchholz, “Origins of the Word Byte,” Annals of the History of Computing 10: 4 (1989): 340.

60. Gardner Hendrie，“从第一台 16 位小型计算机到容错计算机”，计算机博物馆报告（1986 年春季）：6-9。

60. Gardner Hendrie, “From the First 16-bit Mini to Fault-Tolerant Computers,” Computer Museum Report (Spring 1986): 6–9.

61. Arthur Norberg、Judy O'Neill 和 Kerry Freedman，《国防高级研究计划局信息处理技术办公室史》（明尼阿波利斯，明尼苏达州：查尔斯·巴贝奇研究所，1992 年）。

61. Arthur Norberg, Judy O’Neill, and Kerry Freedman, “A History of the Information Processing Techniques Office of the Defense Advanced Research Projects Agency” (Minneapolis, MN: Charles Babbage Institute, 1992).

62. Siewiorek，Bell 和 Newell，《计算机结构》，第 24 章。

62. Siewiorek, Bell, and Newell, Computer Structures, chapter 24.

63. Adele Goldberg 编，《个人工作站的历史》（马萨诸塞州雷丁：Addison-Wesley，1988 年）：151；另见 Siewiorek、Bell 和 Newell 编，《计算机结构》，396-397。

63. Adele Goldberg, ed., A History of Personal Workstations (Reading, MA: Addison-Wesley, 1988): 151; also Siewiorek, Bell, and Newell, Computer Structures, 396–397.

64. Goldberg，《个人工作站的历史》，150–151。

64. Goldberg, History of Personal Workstations, 150–151.

65.截至撰写本文时，史密森尼学会还不属于收藏 IMP 的博物馆之列。

65. As of this writing, the Smithsonian Institution is not among the museums that has collected an IMP.

66.格伦·里夫金和乔治·哈拉尔，《终极企业家：肯·奥尔森和数字设备公司的故事》（芝加哥：当代图书，1988 年）：86-92。

66. Glenn Rifkin and George Harrar, The Ultimate Entrepreneur: the Story of Ken Olsen and Digital Equipment Corporation (Chicago: Contemporary Books, 1988): 86–92.

67.汤姆·沃尔夫，《罗伯特·诺伊斯的修补》《时尚先生》（1983 年 12 月）：356。

67. Tom Wolfe, “The Tinkerings of Robert Noyce,” Esquire (December 1983): 356.

68. Fred Brooks，《Keynotestext1 演讲：语言设计即设计》，收录于 Thomas J. Bergin 和 Richard G. Gibson 编的《编程语言历史-II》（马萨诸塞州雷丁：Addison-Wesley，1996 年）：第 4-16 页。另一台“优雅”电脑 Apple II 的设计师 Steve Wozniak 也承认 Nova 对其产生了影响。Radio Shack TRS-80 Model 100 和 IBM 7090 也被视为“优雅”电脑；但很少有其他电脑被如此评价。

68. Fred Brooks, “Keynotestext1 Address: Language Design as Design,” in Thomas J. Bergin and Richard G. Gibson, eds., History of Programming Languages–-II (Reading, MA: Addison-Wesley, 1996): 4–16. Steve Wozniak, designer of another “elegant” computer, the Apple II, also acknowledged the Nova as an influence. The Radio Shack TRS-80 Model 100, and the IBM 7090 are also regarded as “elegant”; but few other computers are spoken of that way.

69. Michael Hord，《伊利亚克 IV：第一台超级计算机》（马里兰州罗克维尔：计算机科学出版社，1982 年）。值得注意的是，电影《2001：太空漫游》中著名的计算机 HAL 就是在伊利诺伊州厄巴纳“诞生”的。这部电影是在伊利亚克 IV 建造时拍摄的。剧本作者亚瑟·克拉克后来声称，他之所以选择厄巴纳，是因为他的一位教授从英国搬到了那里。

69. Michael Hord, The Illiac IV: the First Supercomputer (Rockville, MD: Computer Science Press, 1982). It is worth noting that HAL, the famous computer of the film 2001: a Space Odyssey, was “born” in Urbana, Illinois. That film was being made as the Illiac IV was being built. Arthur C. Clarke, the author of the screenplay, later claimed that he chose Urbana because one of his professors had moved there from England.

70. “正在进行的革命：英特尔迄今为止的历史”，小册子（加利福尼亚州圣克拉拉：英特尔公司，1984 年）。

70. “A Revolution in Progress: a History of Intel to Date,” brochure (Santa Clara, CA: Intel Corporation, 1984).

71.同上。

71. Ibid.

72.其中包括两家比 DEC 规模大得多的公司的产品：洛克希德和惠普的两个部门都在提供 16 位小型计算机，还有一家初创公司 Interdata，后来被军事承包商珀金埃尔默收购。

72. These included offerings from two companies much larger than DEC: Divisions of both Lockheed and Hewlett-Packard were offering 16-bit minicomputers, as was a start-up, Interdata, that was later purchased by the military contractor Perkin-Elmer.

73. C. Gordon Bell，作者访谈，1992 年 6 月 16 日，加利福尼亚州洛斯阿尔托斯；另见数字设备公司 1980 年 5 月的新闻稿“Decworld”，PDP-11 特别版，副本在作者的收藏中。

73. C. Gordon Bell, interview with author, 16 June 1992, Los Altos, CA; also “Decworld,” newsletter from Digital Equipment Corporation, May 1980, Special Edition on the PDP-11, copy in the author’s collection.

74.哈佛大学计算实验室，《自动序列控制计算器操作手册》，1946年重印版（剑桥：麻省理工学院出版社，1985年），第2章。

74. Harvard University Computation Laboratory, A Manual of Operation for the Automatic Sequence Controlled Calculator, reprint of 1946 edition (Cambridge: MIT Press, 1985), chapter 2.

75.数字设备公司，《PDP-11 处理器手册》（马萨诸塞州梅纳德：数字设备公司，1981），第 2 章；根据 Braun 和 McDonald 的说法，德州仪器已获得总线架构概念的专利；参见 Braun 和 Macdonald 的《微型革命》 109 页。

75. Digital Equipment Corporation, PDP-11 Processor Handbook (Maynard, MA: Digital Equipment Corporation, 1981), chapter 2; according to Braun and McDonald, Texas Instruments had taken out patents on the concept of a bus architecture; see Braun and Macdonald, Revolution in Miniature 109.

76.基于 Intel 8086 系列微处理器构建的个人计算机的数据总线不如 PDP-11 的通用。它们在其他方面与 PDP-11 相似。摩托罗拉 68000 系列微处理器更接近 PDP-11 的模式。

76. Personal computers built around the Intel 8086-series of microprocessors have data buses that are not as general as the PDP-11’s. They do resemble the PDP-11 in other ways. The Motorola 68000 series of microprocessors was more closely patterned on the PDP-11.

77. Jamie Pearson，《数字化工作》（马萨诸塞州贝德福德：Digital Press，1992）：47、59、67。

77. Jamie Pearson, Digital at Work (Bedford, MA: Digital Press, 1992): 47, 59, 67.

78.数字设备公司，《PDP-11 处理器手册》，（马萨诸塞州梅纳德：数字设备公司，1981）：v.

78. Digital Equipment Corporation, PDP-11 Processor Handbook, (Maynard, MA: Digital Equipment Corporation, 1981): v.

79.迪克·鲁宾斯坦，1993 年 2 月 5 日与作者进行电话采访。

79. Dick Rubenstein, telephone interview with the author, February 5, 1993.

80. C. Gordon Bell、J. Craig Mudge 和 John E. McNamara，《计算机工程：DEC 对硬件系统设计的看法》 (马萨诸塞州贝德福德：Digital Press，1978)：383。

80. C. Gordon Bell, J. Craig Mudge, and John E. McNamara, Computer Engineering: a DEC View of Hardware Systems Design (Bedford, MA: Digital Press, 1978): 383.

81. James W. Cortada，《数据处理历史词典：技术》（康涅狄格州韦斯特波特：格林伍德出版社，1987 年）：142。

81. James W. Cortada, Historical Dictionary of Data Processing: Technology (Westport, CT: Greenwood Press, 1987): 142.

82. Pugh 等人，IBM 的 360 和早期 370 系统，第 9 章。

82. Pugh et al., IBM’s 360 and Early 370 Systems, chapter 9.

83. Jim Geraghty，《CICS 概念和用途》（纽约：McGraw-Hill，1994 年）。

83. Jim Geraghty, CICS Concepts and Uses (New York: McGraw-Hill, 1994).

84. Saul Rosen，“PUFFT——普渡大学快速 Fortran 翻译器”，载于 Saul Rosen 主编的《编程系统和语言》（纽约： McGraw-Hill，1967 年）：253-263 页；另见《计算机科学百科全书》第 3 版（纽约： McGraw-Hill，1993 年）：768 页。

84. Saul Rosen, “PUFFT—the Purdue University Fast Fortran Translator,” in Saul Rosen, ed., Programming Systems and Languages (New York: McGraw-Hill, 1967): 253–263; also Encyclopedia of Computer Science, 3rd ed. (New York: McGraw-Hill, 1993): 768.

85. “25周年纪念刊”，滑铁卢大学，计算机服务系通讯（1982年10月）：2。

85. “25th Anniversary Issue,” University of Waterloo, Department of Computing Services Newsletter (October 1982): 2.

86. Ray Argyle，“行业概况……滑铁卢大学的 Wes Graham”，计算机数据：加拿大计算机杂志（1976 年 5 月）：29-30。

86. Ray Argyle, “Industry Profile ... Wes Graham of Waterloo U,” Computer Data: the Canadian Computer Magazine (May 1976): 29–30.

87. Paul Cress、Paul Dirkson 和 J. Wesley Graham，《Fortran IV 与 WATFOR 和 WATFIV》（新泽西州 Englewood Cliffs：Prentice Hall，1970 年）。

87. Paul Cress, Paul Dirkson, and J. Wesley Graham, Fortran IV with WATFOR and WATFIV (Englewood Cliffs, NJ: Prentice Hall, 1970).

88. John G. Kemeny，《人与计算机》（纽约：Scribner's，1972 年）：vii。

88. John G. Kemeny, Man and the Computer (New York: Scribner’s, 1972): vii.

89. Thomas E. Kurtz，“BASIC”，载于 Richard Wexelblat 编，《编程语言史》（纽约：Academic Press，1981 年）：518-519。

89. Thomas E. Kurtz, “BASIC,” in Richard Wexelblat, ed., History of Programming Languages (New York: Academic Press, 1981): 518–519.

90.这个名字的选择是显而易见的；然而，它也是“初学者万能符号指令代码”的首字母缩写。因此它全部用大写字母书写。

90. The choice of the name is obvious; however, it is also an acronym for “Beginner’s All-purpose Symbolic Instruction Code.” Hence it is written in all capitals.

91. William Aspray 和 Bernard O. Williams，“武装美国科学家：国家科学基金会和为大学提供科学计算设施，1950-1973”，计算史年鉴16：4（1994）：60-74。

91. William Aspray and Bernard O. Williams, “Arming American Scientists: NSF and the Provision of Scientific Computing Facilities for Universities, 1950–1973,” Annals of the History of Computing 16: 4 (1994): 60–74.

92.马克·布拉姆霍尔，1997 年 5 月 10 日与作者进行电话采访。

92. Mark Bramhall, telephone interview with the author, 10 May, 1997.

93.一些被抛弃的“神圣”原则包括赋值语句中必须使用“Let”；一行只能有一个语句；不允许语句超出一行。Kemeny 后来开发了“True Basic”，让该语言回归其纯粹的根源，但它从未流行起来。

93. Some of the “sacred” principles abandoned were the mandatory use of “Let” in an assignment statement; having only one statement on a line; and not allowing a statement to continue beyond a single line. Kemeny later developed “True Basic” to return the language to its pure roots, but it never caught on.

第7章

Chapter 7

本章的部分内容首次出现在《历史与技术》 13=1（1996）：1-32。已获许可使用。

Portions of this chapter first appeared in History and Technology, 13=1 (1996): 1– 32. Used with permission.

1.斯图尔特·布兰德，《太空战争：电脑迷的狂热生活和象征性死亡》，《滚石》（1972 年 12 月 7 日）：第 50-58 页。

1. Stewart Brand, “Spacewar: Fanatic Life and Symbolic Death Among the Computer Bums,” Rolling Stone (Dec. 7, 1972): 50–58.

2.道格拉斯·恩格尔巴特（Douglas Engelbart），阿黛尔·戈德堡（Adele Goldberg）编，《个人工作站的历史》（马萨诸塞州雷丁：Addison-Wesley，1988 年）：187。

2. Douglas Engelbart, in Adele Goldberg, ed., History of Personal Workstations (Reading, MA: Addison-Wesley, 1988): 187.

3.致谢“马的故事”，载于马克·吐温的《神秘的陌生人和其他故事》（纽约：哈珀，1922 年）：143–144。

3. Acknowledgements to “A Horse’s Tale,” in Mark Twain, The Mysterious Stranger and Other Stories (New York: Harper, 1922): 143–144.

4.以下对 PDP-10 的描述主要取自 C. Gordon Bell、J. Craig Mudge 和 John McNamara 所著《计算机工程：DEC 对硬件系统设计的视角》（马萨诸塞州贝德福德：Digital Press，1978 年），第 21 章。

4. The following description of the PDP-10 is taken mainly from C. Gordon Bell, J. Craig Mudge, and John McNamara, Computer Engineering: a DEC View of Hardware Systems Design (Bedford, MA: Digital Press, 1978), chapter 21.

5.例如，参见Datamation当时的广告。Digital 的一位发言人称 PDP-6 是“第一台可以称为‘个人’大型机的机器。” Jamie Pearson，《Digital at Work》（马萨诸塞州贝德福德：Digital Press，1992 年）：54–55。

5. See, for example, advertisements in Datamation from that period. A Digital spokesperson called the PDP-6 “the first of what might be called a ‘personal’ mainframe.” Jamie Pearson, Digital at Work (Bedford, MA: Digital Press, 1992): 54–55.

6. 1972 年中期推出的型号使用集成电路进行逻辑处理，但内存仍然采用磁芯实现。

6. Models introduced in mid-1972 used integrated circuits for logic, but memory was still implemented with magnetic core.

7. PDP-6 于 1964 年推出，PDP-10 于 1967 年推出，DEC-System 10 于 1971 年推出。共售出 23 台 PDP-6。到 1976 年，已安装约 450 台 PDP-10 和 DEC-System 10。

7. The PDP-6 was introduced in 1964, the PDP-10 in 1967, and the DEC-System 10 in 1971. 23 PDP-6s were sold. By 1976 around 450 PDP-10s and DEC-System 10s had been installed.

8. Bill Gosper，引自 Steven Levy 所著《黑客：计算机革命的英雄》（纽约：Anchor Doubleday，1984 年）：第 67 页。

8. Bill Gosper, quoted in Steven Levy, Hackers: Heroes of the Computer Revolution (New York: Anchor Doubleday, 1984): 67.

9.爱德华·弗雷德金（Edward Fredkin）对作者的采访，1993 年 5 月 21 日。

9. Edward Fredkin interview with the author, 21 May, 1993.

10. Bell 等人，《计算机工程》，第 507 页的表格。

10. Bell et al., Computer Engineering, table on p. 507.

11.数字设备公司，《TOPS-10 技术摘要》 (Maynard, MA, nd)。波士顿计算机博物馆，PDP-10 档案。

11. Digital Equipment Corporation, “TOPS-10 Technical Summary” (Maynard, MA, n.d.). Computer Museum, Boston, PDP-10 archives.

12.波士顿计算机博物馆，PDP-10 档案，盒子 A-242。

12. Computer Museum, Boston, PDP-10 archives, box A-242.

13.剑桥咨询公司 Bolt Beranek and Newman (BBN) 还为高级研究计划局开发了另一个适用于 PDP-10 的操作系统。“TENEX”比 TOPS-10 更加完善。

13. The Cambridge consulting firm Bolt Beranek and Newman (BBN) also developed another operating system for the PDP-10, for the Advanced Research Projects Agency. “TENEX” was more polished than TOPS-10.

14. David Ahl，《电脑游戏》，Anthony Ralston 和 Edwin Reilly 编，《计算机科学百科全书》，第三版（纽约：Van Nostrand，1993 年）：285-287。

14. David Ahl, “Computer Games,” in Anthony Ralston and Edwin Reilly, eds., Encyclopedia of Computer Science, third edition (New York: Van Nostrand, 1993): 285–287.

15. MAXC 与 DP-10 软件兼容，但使用半导体存储器代替核心。参见 Douglas Smith 等人的《摸索未来：施乐公司如何发明并忽视了第一台个人电脑》（纽约：William Morrow，1988 年）：第 144-145 页；另见 Tekla Perry 的《PARC 内部：‘信息架构师’》IEEE Spectrum（1985 年 10 月）：第 62-75 页。

15. The MAXC was software-compatible with a DP-10 but used semiconductor memory instead of core. See Douglas Smith et al., Fumbling the Future: How Xerox Invented, Then Ignored, the First Personal Computer (New York: William Morrow, 1988): 144–145; also Tekla Perry, “Inside the PARC: the ‘Information Architects’ ” IEEE Spectrum (October 1985): 62–75.

16.马克斯·帕莱夫斯基 (Max Palevsky) 的证词，1974 年 7 月 10 日，美国诉 IBM，加入号 1912，盒子 31，记录稿；哈格利博物馆档案馆。

16. Deposition by Max Palevsky, 10 July 1974, U.S. v. IBM, Accession 1912, Box 31, transcript; Hagley Museum Archives.

17.其中一个演变成了 Compuserve 网络。截至撰写本文时，Compuserve 用户仍通过八进制表示法来标识；该表示法仅使用 0 到 7 的数字。这反映了它从 PDP-10 操作系统的演化，后者也使用八进制而不是十进制数字。

17. One of them evolved into the Compuserve network. As of this writing, Compuserve subscribers are still identified by a number written in the octal notation; which uses only the digits zero through seven. That reflects its descent from PDP-10 operating systems, which also used octal rather than decimal numbers.

18. Mark Halpern，《获得资助的梦想》，计算机史年鉴16=3（1994 年）：61-64。

18. Mark Halpern, “Dreams That Get Funded,” Annals of the History of Computing 16=3 (1994): 61–64.

19.史蒂芬·曼尼斯 (Stephen Manes) 和保罗·安德鲁斯 (Paul Andrews)，《盖茨：微软大亨如何重塑一个行业并成为美国首富》 (纽约：Doubleday，1993 年）：第 28-36 页。

19. Stephen Manes and Paul Andrews, Gates: How Microsoft’s Mogul Reinvented an Industry, and Made Himself the Richest Man in America (New York: Doubleday, 1993): 28–36.

20. Brand，《太空战争：狂热生活》；另请参阅Freiberger，《硅谷之火》，第 100–102 页。

20. Brand, “Spacewar: The Fanatic Life”; also Freiberger, Fire in the Valley, 100–102.

21. Chuck House，《惠普和个人计算系统》，载于 Adele Goldberg 主编的《个人工作站的历史》（马萨诸塞州雷丁：Addison-Wesley，1988 年）：413-414；C. Gordon Bell，作者访谈，1992 年 6 月 16 日。

21. Chuck House, “Hewlett-Packard and Personal Computing Systems,” in Adele Goldberg, ed., History of Personal Workstations (Reading, MA: Addison-Wesley, 1988): 413–414; C. Gordon Bell, interview with the author, 16 June, 1992.

22. Wesley A. Clark，《LINC 成立较早，规模较小》，载 Goldberg 著《个人工作站历史》，第 347–391 页。

22. Wesley A. Clark, “The LINC was Early and Small,” in Goldberg, History of Personal Workstations, 347–391.

23.培生，《数字化工作》，52。

23. Pearson, Digital at Work, 52.

24.另一个领域是计算机游戏。不过，我不讨论这些领域，主要是因为我觉得它们的客户，尽管作为半导体的大众市场很重要，但并不像计算器用户那样是关键群体。

24. Another place was in computer games. I do not discuss these, however, mainly because I feel their customers, however important as a mass market for semiconductors, were not the critical community that calculator users were.

25. Peggy A. Kidwell 和 Paul Ceruzzi，《数字计算的里程碑：史密森尼图画史》（华盛顿特区：史密森尼出版社，1994 年）；Edwin Darby，《一切都加起来：Victor Comptometer 公司的成长》（芝加哥：Victor Comptometer 公司，1968 年）。

25. Peggy A. Kidwell and Paul Ceruzzi, Landmarks in Digital Computing: a Smithsonian Pictorial History (Washington, DC: Smithsonian Press, 1994); Edwin Darby, It All Adds Up: the Growth of Victor Comptometer Corporation (Chicago: Victor Comptometer Corp., 1968).

26. An Wang，《Lessons》（马萨诸塞州雷丁：Addison-Wesley，1986 年）：126–159。王安的突破之道在于他能够仅使用几百个晶体管实现所有功能。他利用了霍华德·艾肯及其哈佛团队在 20 世纪 40 年代利用 Harvard Mark I 开发的技巧。

26. An Wang, Lessons (Reading, MA: Addison-Wesley, 1986): 126–159. The key to Wang’s breakthrough was his ability to perform all the functions using only a few hundred transistors. He did this by exploiting tricks that Howard Aiken and his Harvard team had developed in the 1940s with the Harvard Mark I.

27. Chuck House，《惠普与个人计算系统》，第 401-427 页；另见《Olivetti Programma 101 台式计算器》，收录于 Bell and Newell 的《计算机结构》，第 237-242 页。

27. Chuck House, “Hewlett-Packard and Personal Computing Systems,” 401– 427; also “The Olivetti Programma 101 Desk Calculator,” in Bell and Newell, Computer Structures, 237–242.

28.计算器 IC 的主要供应商是德州仪器和罗克韦尔，后者的 Autonetics 部门制造了 Minuteman 制导系统。使用其芯片的计算器以各种名称出售，尽管最终 TI 和罗克韦尔以自己的名称进入市场。罗克韦尔后来又回到了仅作为芯片供应商的地位。

28. The chief suppliers of ICs for calculators were Texas Instruments and Rockwell, whose Autonetics Division built the Minuteman Guidance systems. Calculators using their chips were sold under a variety of names, although eventually TI and Rockwell entered the market with machines under their own names. Rockwell later returned to being only a chip supplier.

29.美国国家历史博物馆，计算器收藏。百老汇热门剧作《明年同一时间》甚至提到了 Bowmar，该剧讲述了一位会计师与一位女性一年一度的恋情。这位会计师（电影版中由艾伦·阿尔达饰演）使用 Bowmar 来记录这段恋情。

29. National Museum of American History, calculator collections. A Bowmar was even mentioned in the hit Broadway play “Same Time Next Year,” which was about an accountant’s once-a-year affair with a woman. The accountant (played by Alan Alda in the movie version) used the Bowmar to keep track of the affair.

30. Electronics (1980 年 4 月 17 日)：第 397-398 页；另见 Rockwell International 出版的小册子“20 世纪 70 年代的微电子学”，1974 年，第 39 页。

30. Electronics (April 17, 1980): 397–398; also “Microelectronics in the 1970s,” booklet from Rockwell International, 1974, 39.

31.美国最受欢迎的工程师计算尺制造商 Keufell & Esser 于 1975 年停止生产。在接下来的大约 10 年里，其他公司继续生产仅具有基本算术和对数刻度的廉价计算尺。

31. Keufell & Esser, the U.S. manufacturer of the most popular engineer’s slide rules, stopped production in 1975. For about the next decade, other companies continued to make cheaper slide rules that had only the basic arithmetic and log scales.

32. Chung C. Tung，“‘个人计算机’：一款完全可编程的袖珍计算器”，《惠普杂志》（1974 年）：2-7。

32. Chung C. Tung, “The ‘Personal Computer’: a Fully Programmable Pocket Calculator,” Hewlett-Packard Journal (1974): 2–7.

33. Gordon Moore，“微处理器和集成电子技术”，IEEE 64 会议纪要（1976 年 6 月）：837-841。

33. Gordon Moore, “Microprocessors and Integrated Electronics Technology,” Proceedings of the IEEE 64 (June 1976): 837–841.

34. Joseph Weizenbaum，《计算机能力与人类理性：从判断到计算》（旧金山：WH Freeman，1976 年），第 4 章。

34. Joseph Weizenbaum, Computer Power and Human Reason: From Judgment to Calculation (San Francisco: W. H. Freeman, 1976), chapter 4.

35. 65-notestext1s（HP-65 用户俱乐部简讯）2：1（1975 年 1 月）：7。HP-65 的客户绝大多数是男性；该简讯特别记录了第一位加入用户俱乐部的女性会员，当时俱乐部成立一年后。

35. 65-notestext1s (Newsletter of the HP-65 Users’ Club) 2: 1 (January 1975): 7. HP-65 customers were overwhelmingly male; the newsletter made a special notestext1 of the first female member to join the users club, a year after its founding.

36.Weizenbaum，《计算机能力》，116。

36. Weizenbaum, Computer Power, 116.

37. Paul Freiberger，《硅谷之火：个人电脑的诞生》（加州伯克利：Oxborne=McGraw-Hill，1984 年）。

37. Paul Freiberger, Fire in the Valley: the Making of the Personal Computer (Berkeley, CA: Oxborne=McGraw-Hill, 1984).

38.除了上文引用的“HP-65 notestext1s”中的常规专栏外，作者还在德州仪器用户俱乐部通讯以及 20 世纪 70 年代末在德国出版的针对计算器所有者的通讯“Display”中发现了类似的比较。

38. In addition to a regular column that appeared in “HP-65 notestext1s,” cited above, the author has found similar comparisons in a Texas Instruments users club newsletter, as well as in “Display,” a newsletter for calculator owners published in Germany in the late 1970s.

39. Ted Nelson，《计算机自由》（印第安纳州南本德：Ted Nelson，1974 年）。

39. Ted Nelson, Computer Lib (South Bend, IN: Ted Nelson, 1974).

40. “冯·诺依曼”论证的依据是，与通用计算机不同，大多数计算器将程序存储在与数据分开的内存中。事实上，程序与数据存储在同一芯片上，但计算器制造商竖起了一道“墙”来防止两者相遇。这样做是为了让非专业人士更容易使用机器。公共内存通常被认为是真正计算机的核心定义特征。大多数可编程计算器都具有的另一个特性是“条件分支”：能够根据上一次计算的结果选择替代指令序列。这是计算机时代之前机器所缺乏的特性：哈佛 Mark I、早期贝尔实验室中继计算机和早期 Zuse 计算机。

40. The “von Neumann” argument came from the fact that most calculators, unlike general-purpose computers, stored their programs in a memory deliberately kept separate from data. In fact, the program was stored on the same chips as the data, but the calculator manufacturers erected a “wall” to prevent the twain from meeting. This was done to make the machine easier to use by nonspecialists. A common memory is often regarded as a central defining feature of a true computer. Another property, which most programmable calculators did have, was “conditional branching”: the ability to select alternate sequences of instructions based on the results of a previous calculation. That was a property lacking in the machines of the immediate precomputer era: the Harvard Mark I, the early Bell Labs relay computers, and the early Zuse computers.

41. “可编程袖珍计算器的拥有者：他以为他是谁？” HP-65 notestext1s 3: 6 (1976): 2。

41. “The Programmable Pocket Calculator Owner: Who Does He Think He Is?” HP-65 notestext1s 3: 6 (1976): 2.

42. HP-65 notes text1s 2: 1 (1975): 4–7。

42. HP-65 notestext1s 2: 1 (1975): 4–7.

43. Gordon E. Moore，《数字集成电子学的进展》，《国际电子设备会议论文集》（1975 年 12 月）：第 11-13 页。Robert Noyce 表示，Moore 于 1964 年首次注意到这一趋势：Noyce，《微电子学》，《科学美国人》（1977 年 9 月）：第 63-69 页。Moore 预测，到 1980 年，这一速度将趋于平稳，每两年翻一番。这导致大众媒体对“摩尔定律”的确切含义产生了混淆。Bell、Mudge 和 MacNamara（《计算机结构》，第 90 页）表示，该定律从 1958 年到 1972 年每年翻一番，此后每 18 个月翻一番。从 20 世纪 70 年代到本文撰写之时，内存芯片密度每 18 个月翻一番。

43. Gordon E. Moore, “Progress in Digital Integrated Electronics,” Proceedings International Electron Devices Meeting (December 1975): 11–13. Robert Noyce stated that Moore first noticed this trend in 1964: Noyce, “Microelectronics,” Scientific American (September 1977): 63–69. Moore predicted that the rate would flatten out to a doubling every two years by 1980. That has led to confusion in the popular press over what exactly is meant by “Moore’s Law.” Bell, Mudge, and MacNamara (Computer Structures, 90) state the law as doubling every year from 1958 until 1972, then every eighteen months thereafter. Memory chip density, from the 1970s to the time of this writing, has been doubling every eighteen months.

44. Clifford Barney，《他从零开始创立了 MOS》，《电子周刊》（1984 年 10 月 8 日）：第 64 页。

44. Clifford Barney, “He Started MOS From Scratch,” Electronics Week (October 8, 1984): 64.

45.霍夫回忆说，阿迪·J·坎巴塔 1969 年出版的《大规模集成电路简介》一书影响巨大。这本书描述了亨利·福特和他的 T 型车所面临的困境的现代版本：正是大规模生产技术使 T 型车成为一款高品质、低价位的汽车，而随着市场的发展，福特很难甚至不可能改变 T 型车的设计。

45. Hoff recalls a book by Adi J. Khambata, Introduction to LSI, published in 1969, as very influential. The book gave modern version of the dilemma faced by Henry Ford and his Model T: the very same mass-production techniques that made the Model T a high-quality, low-priced car made it difficult if not impossible for Ford to change the Model T’s design as the market evolved.

46. Trudy E. Bell，“20 年后获得专利：微处理器的真正父亲”，The Institute (IEEE) 14: 10 (1990 年 11 月)：1；另请参阅美国国家历史博物馆电力部门的策展文件，德州仪器收藏；另请参阅 Don Clark，“芯片专利的高风险战争”，旧金山纪事报(1990 年 9 月 8 日)：b1–b3；另请参阅 Michael Antonof，“Gilbert Who?” Popular Science (1991 年 2 月)：70–73。

46. Trudy E. Bell, “Patented 20 Years Later: the Microprocessor’s True Father,” The Institute (IEEE) 14: 10 (November 1990): 1; also National Museum of American History, Division of Electricity, curatorial files, Texas Instruments collection; also Don Clark, “High-Stakes War Over Chip Patents,” San Francisco Chronicle (September 8, 1990): b1–b3; also Michael Antonof, “Gilbert Who?” Popular Science (February 1991): 70–73.

47.例如，请参阅 Robert Noyce 和 Marcian Hoff 的《英特尔微处理器设计史》；IEEE Micro 1（1981 年 2 月）：8-22。

47. See for example Robert Noyce and Marcian Hoff, “A History of Microprocessor Design at Intel”; IEEE Micro 1 (February 1981): 8–22.

48. Kenneth A. Brown，与 Hoff 的访谈，摘自 Brown 著《发明家在工作》（华盛顿州雷德蒙德：Tempus Books）：283–307。

48. Kenneth A. Brown, interview with Hoff, in Brown, Inventors at Work (Redmond, WA: Tempus Books): 283–307.

49. William Barden Jr.，《如何购买和使用小型计算机和微型计算机》（印第安纳波利斯：Howard Sams，1976 年）：101-103。

49. William Barden Jr., How to Buy and Use Minicomputers and Microcomputers (Indianapolis: Howard Sams, 1976): 101–103.

50.英特尔公司，企业传播部，《正在进行的革命：英特尔迄今为止的历史》 (加利福尼亚州圣克拉拉：英特尔，1984 年)：第 12 页。

50. Intel Corporation, Corporate Communications Department, “A Revolution in Progress: a History of Intel to Date” (Santa Clara, CA: Intel, 1984): 12.

51. 《电子新闻》（1971年11月15日）。

51. Electronic News (November 15, 1971).

52.英特尔，《正在进行的革命》，第 21 页。

52. Intel, “A Revolution in Progress,” 21.

53. Elvia Faggin，《Faggin 对第一台微处理器的贡献》，致编辑的信，《圣何塞水星报》（1986 年 10 月 3 日）：6b；Marcian Hoff 的回复，《专利并不能说明微处理器的全部故事》，同上。（1986 年 10 月 12 日）：10b；另见《如果 Hyatt 没有发明微处理器，那是谁发明的？》同上。（1990 年 12 月 2 日）：27。

53. Elvia Faggin, “Faggin Contributed to First Microprocessor,” letter to the Editor, San Jose Mercury News (October 3, 1986): 6b; reply by Marcian Hoff, “Patents Don’t Tell Whole Microprocessor Tale,” ibid. (October 12, 1986): 10b; also “If Hyatt Didn’t Invent the Microprocessor, Who Did?” ibid. (December 2, 1990): 27.

54. Hoff，“专利并不能说明微处理器的全部故事”，106。

54. Hoff, “Patents Don’t Tell Whole Microprocessor Tale,” 106.

55. Noyce 和 Hoff，《微处理器设计史》；Lamont Wood，《发明个人电脑的人》，《美国发明与技术遗产》（1994 年秋季）：第 64 页。

55. Noyce and Hoff, “A History of Microprocessor Design”; also Lamont Wood, “The Man Who Invented the PC,” American Heritage of Invention & Technology (Fall 1994): 64.

56.英特尔，《正在进行的革命》，第 14 页。

56. Intel, “A Revolution in Progress,” 14.

57. Noyce 和 Hoff，“微处理器发展史”。

57. Noyce and Hoff, “A History of Microprocessor Development.”

58.计算机博物馆报告17 (1986 年秋季)：10-11。

58. Computer Museum Report 17 (Fall 1986): 10–11.

59.英特尔公司，《正在进行的革命》，第 13 页。

59. Intel Corporation, “A Revolution in Progress,” 13.

60. Robert Slater，《硅中的肖像》（剑桥：麻省理工学院出版社，1992 年）：251-261。

60. Robert Slater, Portraits in Silicon (Cambridge: MIT Press, 1992): 251–261.

61. Noyce 和 Hoff，《英特尔微处理器设计史》，第 14 页。

61. Noyce and Hoff, “A History of Microprocessor Design at Intel,” 14.

62.此声明基于与几位参与早期微处理器开发的英特尔员工的对话，其中包括 Ted Hoff 和 John Wharton。英特尔系统用于在 1976 年夏季奥运会期间保持得分。那一年，Nadia Comaneci 在体操比赛中获得了满分“10”，但系统无法显示这个分数，因为它没有被编程为显示“9.99”以上的任何分数。然而，这一限制与英特尔系统的字长比它所取代的微型计算机短这一事实无关。

62. This statement is based on conversations with several Intel employees who were involved with early microprocessor development, including Ted Hoff and John Wharton. Intel systems were used to keep scores during the 1976 Summer Olympics. That was the year Nadia Comaneci received a perfect “10” in gymnastics, a score that the system was unable to display, as it had not been programmed to display anything over “9.99.” That limit, however, had nothing to do with the fact that the Intel systems had a shorter word length than the minicomputers it replaced.

63. Susan Douglas，“技术的对立使用与企业竞争：以无线电广播为例”，William Aspray 主编，《技术竞争力》（纽约：IEEE，1993 年）：第 208-219 页。

63. Susan Douglas, “Oppositional Uses of Technology and Corporate Competition: the Case of Radio Broadcasting,” in William Aspray, ed., Technological Competitiveness (New York: IEEE, 1993): 208–219.

64.世贸中心的建设摧毁了无线电街，但那时集成电子技术已在顺利发展。单个微处理器所含的电路可能比无线电街上所有商店的全部商品还要多。

64. The construction of the World Trade Center obliterated Radio Row, but by then integrated electronics was well underway. A single microprocessor might contain more circuits than the entire contents of every store on Radio Row.

65.QST （1974年3 月）：154。

65. QST (March 1974): 154.

66. Stan Veit，《Stan Veit 的个人电脑史》（北卡罗来纳州阿什维尔：World-Comm，1993 年）：11；另请参阅 Thomas Haddock，《个人电脑收藏家指南》（阿拉巴马州佛罗伦萨：Thomas Haddock，1993 年）：20。

66. Stan Veit, Stan Veit’s History of the Personal Computer (Asheville, NC: World-Comm, 1993): 11; also Thomas Haddock, A Collector’s Guide to Personal Computers (Florence, AL: Thomas Haddock, 1993): 20.

67. “构建 Mark-8，你的个人小型计算机”，《无线电电子学》（1974 年 7 月）：封面，第 29-33 页。

67. “Build the Mark-8, Your Personal Minicomputer,” Radio-Electronics (July 1974): cover, 29–33.

68.同上。用户俱乐部变成了 Digital Group，这是接下来几年个人计算机领域一家颇具影响力的公司。参见 Jonathan Titus，1984 年 6 月 18 日致计算机博物馆的信，波士顿计算机博物馆，个人计算机档案。

68. Ibid. The users club became the Digital Group, an influential company in personal computing for the next several years. See Jonathan Titus, letter to the Computer Museum, June 18, 1984, Computer Museum, Boston, Personal Computer archives.

69. NMAH Collections；另见 Steve Ditlea 编辑的《Digital Deli》（纽约：Workman，1984 年）：37。

69. NMAH Collections; also Steve Ditlea, ed., Digital Deli (New York: Workman, 1984): 37.

70. “建造Mark-8”，第33页。

70. “Build the Mark-8,” 33.

71. Don Lancaster，《电视打字机》，《无线电电子学》（1973 年 9 月）：封面，第 43-52 页；Felsenstein 被引用于《计算机博物馆报告》第 17 期（1986 年秋季）：第 16 页。

71. Don Lancaster, “TV-Typewriter,” Radio-Electronics (September 1973): cover, 43–52; Felsenstein is quoted in the Computer Museum Report 17 (Fall 1986): 16.

72. H. Edward Roberts 和 William Yates，“独家！Altair 8800：有史以来最强大的小型计算机项目——成本不到 400 美元”，《大众电子》（1975 年 1 月）：封面，第 33-38 页。

72. H. Edward Roberts and William Yates, “Exclusive! Altair 8800: the Most Powerful Minicomputer Project Ever Presented—Can be Built for Under $400,” Popular Electronics (January 1975): cover, 33–38.

73. Altair 推出后不久，记者们就开始将这些机器称为“微型计算机”，这是一个准确但又含糊不清的术语，因为它可能意味着两种不同的东西。微型计算机使用微处理器，而小型计算机则不使用微处理器。当时确实如此，尽管后来几乎所有类型的机器都会使用微处理器。另一种定义是微型计算机比小型计算机更小或更便宜。Altair 兼具两者，但其低成本比其小巧更重要。

73. Not long after the Altair’s introduction, journalists began calling these machines “microcomputers,” an accurate but also ambiguous term, as it could imply two different things. A microcomputer used a microprocessor, and minicomputers did not. That was true at the time, although eventually nearly every class of machine would use microprocessors. The other definition was that a microcomputer was smaller and=or cheaper than a minicomputer. The Altair was both, but its low cost was more important than its small size.

74. Intel，《正在进行的革命》，第 14 页；另见 Veit，《斯坦·维特的个人电脑史》，第 43 页；Veit 表示，Roberts 获得的芯片存在外观缺陷，但 Roberts 和 Intel 都明确表示，Altair 中使用的 8080 芯片没有任何缺陷；参见“Computer notes text1s”，MITS 1：3（1975 年 8 月）：2（美国国家历史博物馆，Mims-Altair 文件）。事实上，8080 的制造成本非常低，而且它对 PC 市场的公平市场价格知之甚少。

74. Intel, “A Revolution in Progress,” 14; also Veit, Stan Veit’s History of the Personal Computer, 43; Veit stated that Roberts obtained chips that had cosmetic flaws, but Roberts and Intel both state flatly that the 8080 chips used in the Altair were not defective in any way; see “Computer notestext1s,” MITS 1: 3 (August 1975): 2 (National Museum of American History, Mims-Altair file). The fact was that the 8080 cost Intel very little to manufacture, and it had little sense of what a fair market price to the PC market should be.

75.在《大众电子学》文章中拼写为“buss” 。

75. Spelled “buss” in the Popular Electronics article.

76. Veit 在《Stan Veit 的历史》一书中认为，正是由于 Railway Express 的无能，我们才做出了使用总线的重大决定；其他人则声称，这一决定源于 Roberts 以特别优惠的价格找到了 100 插槽连接器的供应。设计上的改变使 Altair 更像当时的微型计算机，尽管这使其组装起来更加困难。

76. Veit, in Stan Veit’s History, argues that it is to Railway Express’s ineptitude that we owe the momentous decision to have a bus; others claim the decision came from Roberts’s finding a supply of 100-slot connectors at an especially good price. The design change made the Altair more like the minicomputers of the day, though it made it more difficult to assemble.

77. Roberts和Yates，“独家！”34。

77. Roberts and Yates, “Exclusive!” 34.

78.例如请参阅 Steven Manes 和 Paul Andrews，第 64 页。

78. See, for example Steven Manes and Paul Andrews, 64.

79. Jim Warren，“个人计算：计算机专业人员概述”，NCC Proceedings 46（1977）：493-498。

79. Jim Warren, “Personal Computing: an Overview for Computer Professionals,” NCC Proceedings 46 (1977): 493–498.

80.其中包括“多通道数据采集系统”、“机器控制器”、“供暖、空调、除湿自动控制器”以及“机器人的大脑”等。

80. These included “Multichannel data acquisition system,” “Machine controller,” “Automatic controller for heat, air conditioning, dehumidifying,” as well as “Brain for a robot,” and others.

81. Veit，Stan Veit's History，57–64，列出了 IMSAI 和 Altair 之间的主要区别。

81. Veit, Stan Veit’s History, 57–64, gives the main differences between the IMSAI and the Altair.

82.这就是为什么在两种输入输出设备都不再使用之后，TTY（电传打字机）和LPT（行式打印机）这两个缩写词仍然保留在个人计算机的操作系统中。

82. This is the reason that the acronyms TTY for Teletype and LPT for line printer survived into the operating systems of personal computers, long after both input=output devices fell from use.

83. Veit，Stan Veit 的历史。

83. Veit, Stan Veit’s History.

84. Pugh 等人，IBM 的 360 和早期 370 系统(剑桥：麻省理工学院出版社，1991 年）：510-521。

84. Pugh et al., IBM’s 360 and Early 370 Systems (Cambridge: MIT Press, 1991): 510–521.

85. Clifford Barney，《成就奖 [Alan F. Shugart]》，《电子周刊》（1985 年 1 月 14 日）第 40-44 页。

85. Clifford Barney, “Award for Achievement [Alan F. Shugart], ” Electronics Week (January 14, 1985) 40–44.

86. Jon Eklund，《个人计算机》，Anthony Ralston 和 Edwin Reilley 编，《计算机科学百科全书》，第 3 版（纽约：van Nostrand Reinhold，1993 年）：460-463。

86. Jon Eklund, “Personal Computers,” in Anthony Ralston and Edwin Reilley, eds., Encyclopedia of Computer Science, 3rd ed. (New York: van Nostrand Reinhold, 1993): 460–463.

87. Forrest Mims III，《Altair 8800 十周年纪念》，《计算机与电子》（1985 年 1 月）：第 62 页。罗伯特的说法一直受到其他人的质疑，至今仍存在争议。

87. Forrest Mims III, “The Tenth Anniversary of the Altair 8800,” Computers and Electronics (January 1985): 62. Robert’s account has been disputed by others and remains controversial.

88.史蒂芬·曼尼斯和保罗·安德鲁斯，《盖茨：微软大亨如何重塑一个行业并成为美国首富》 (纽约：Doubleday，1993 年）：第 63 页。

88. Stephen Manes and Paul Andrews, Gates: How Microsoft’s Mogul Reinvented an Industry, and Made Himself the Richest Man in America (New York: Doubleday, 1993): 63.

89. MITS 公司，计算机 notestext1s 1: 2（1975 年 7 月）：6-7，美国国家历史博物馆，Altair 文件。

89. MITS Corporation, Computer notestext1s 1: 2 (July 1975): 6–7, National Museum of American History, Altair files.

90.数字设备公司，《编程入门》（马萨诸塞州梅纳德，1972 年）：9=4–9=5。Microsoft BASIC 还与达特茅斯学院不同，它允许一行中有多个语句，将“Let”和“End”设为可选项，并建议程序员“删除所有 REM [注释] 语句……从程序中删除所有不必要的空格”。（MITS Altair BASIC 参考手册，56；美国国家历史博物馆，Altair 馆藏文件。）

90. Digital Equipment Corporation, “Introduction to Programming” (Maynard, MA, 1972): 9=4–9=5. Microsoft BASIC also broke with Dartmouth by allowing multiple statements on a line, by having “Let” and “End” optional, by recommending that a programmer “delete all REM [remark] statements ... delete all unnecessary spaces from your program.” (MITS Altair BASIC Reference Manual, 56; National Museum of American History, Altair Curatorial File.)

91. Manes 和 Andrews，Gates，第 2 章和第 3 章；有关John Norton 的讨论，参见 Billy Goodman 的“Practicing Safe Software”，Air & Space = Smithsonian（1994 年 9 月）：60-67；另参见 Paul Ceruzzi 的Beyond the Limits（剑桥：麻省理工学院出版社，1989 年），第 9 章。

91. Manes and Andrews, Gates, chapters 2 and 3; for a discussion of John Norton, see Billy Goodman, “Practicing Safe Software,” Air & Space=Smithsonian (September 1994): 60–67; also Paul Ceruzzi, Beyond the Limits (Cambridge: MIT Press, 1989), chapter 9.

92.这也是个争议很大的问题。罗伯茨坚持认为 MITS 公司拥有 BASIC 的版权。盖茨在 1976 年 4 月给《计算机笔记文本》简报的一封信中说：“我不是 MITS 的员工”，但那是在他与罗伯茨的关系越来越深之后写的。另见 Stephen Manes 和 Paul Andrews 的《盖茨》。

92. This, too, is a matter of great dispute. Roberts insists that MITS had the rights to BASIC. In a letter to the newsletter “Computer notestext1s” on April 1976, Gates stated, “I am not a MITS employee,” but that was written after his rift with Roberts had grown deep. See also Stephen Manes and Paul Andrews, Gates.

93. MITS 公司，《计算机笔记文本》（1976 年 2 月）：第 3 页。这封公开信分发给许多业余出版物，并被广泛阅读。

93. MITS Corporation, “Computer notestext1s” (February 1976): 3. The open letter was distributed to many hobbyist publications and was widely read.

94. C. Gordon Bell，1992 年 6 月，加利福尼亚州洛斯加托斯，对作者的采访；Mark Bramhall，1997 年 5 月 10 日，对作者的电话采访。

94. C. Gordon Bell, interview with the author, June 1992, Los Gatos, CA; Mark Bramhall, telephone interview with the author, 10 May 1997.

95.例如，该术语从 20 世纪 60 年代末开始用于 IBM System=360；参见 Pugh (1991)，第 6 章。

95. This term had been used, for example, with the IBM System=360 beginning in the late 1960s; see Pugh (1991), chapter 6.

96. C. Gordon Bell，作者访谈。Bell 声称他是 PIP 程序的作者，该程序出现在 CP=M 和 MSDOS 的变体中；他说该名字来自 Edward Fredkin。

96. C. Gordon Bell, interview with the author. Bell stated that he was the author of the PIP program, which found its way onto CP=M and in variations to MSDOS; he says the name came from Edward Fredkin.

97. Pearson，《数字化工作》，64–65、86；C. Gordon Bell 与作者的访谈，1992 年 6 月。

97. Pearson, Digital at Work, 64–65, 86; also C. Gordon Bell, interview with the author, June 1992.

98. Gary Kildall，《微型计算机软件设计——检查点》，全国计算机会议44（1975）：99-106；Kildall 也引用自 Susan Lammers 编的《程序员在工作》（华盛顿州雷德蒙德：微软出版社，1989）：61。

98. Gary Kildall, “Microcomputer Software Design—a Checkpoint,” National Computer Conference 44 (1975): 99–106; also Kildall, quoted in Susan Lammers, ed., Programmers at Work (Redmond, WA: Microsoft Press, 1989): 61.

99. Gary Kildall，《CP=M：8 位和 16 位操作系统系列》，Byte（1981 年 6 月）：216-229。由于 DEC 小型计算机与 8080 微处理器之间存在差异，因此 CP=M 的实际代码不同且完全原创，即使语法和词汇相似。

99. Gary Kildall, “CP=M: A Family of 8-and 16-Bit Operating Systems,” Byte, (June 1981): 216–229. Because of the differences between DEC minicomputers and the 8080 microprocessor, the actual code of CP=M was different and wholly original, even if the syntax and vocabulary were similar.

100.上述论证基于作者所拥有的 PDP-10 和 CP=M 手册，以及与 Kip Crosby 的对话，我很感谢他在互联网讨论论坛上发布了这个问题。

100. The above argument is based on PDP-10 and CP=M manuals in the author’s possession, as well as conversations with Kip Crosby, to whom I am grateful for posting this question over an Internet discussion forum.

101. Jim C. Warren，《软盘操作系统初探》，《Dr. Dobb's Journal》（1976 年 4 月）：5。

101. Jim C. Warren, “First Word on a Floppy-disc Operating System,” Dr. Dobb’s Journal (April 1976): 5.

102. Robert Slater，《硅中的肖像》（剑桥：麻省理工学院出版社，1982 年），第 23 章。

102. Robert Slater, Portraits in Silicon (Cambridge: MIT Press, 1982), chapter 23.

103.同上；另请参阅 Stan Veit的《Stan Veit 的历史》，第 64 页；以及 Digital Research 的《C=M 功能和设施简介》（1976 年），手册归作者所有。

103. Ibid.; also Stan Veit, Stan Veit’s History, 64; and Digital Research, “An Introduction to C=M Features and Facilities” (1976), manual in the author’s possession.

第八章

Chapter 8

1. C. Gordon Bell，1992 年 6 月 16 日对作者的采访，加利福尼亚州洛斯阿尔托斯。

1. C. Gordon Bell, interview with the author, 16 June 1992, Los Altos, California.

2.迪克·鲁宾斯坦（Dick Rubenstein），对作者的采访；还有贝尔的采访。

2. Dick Rubenstein, interview with the author; also Bell, interview.

3.贝尔，采访。

3. Bell, interview.

4. C. Gordon Bell 等人，《计算机工程：DEC 对硬件系统设计的视角》（马萨诸塞州贝德福德：Digital Press，1978 年），图表见第 195 页。

4. C. Gordon Bell et al., Computer Engineering: a DEC View of Hardware Systems Design (Bedford, MA: Digital Press, 1978), graph on page 195.

5.同上，第13页。

5. Ibid., 13.

6.珀金埃尔默后来因其在哈勃太空望远镜镜子制造中所发挥的作用而闻名；但它因制造芯片所用光掩模所需的关键光学组件而闻名。

6. Perkin-Elmer later became known for its role in building the Hubble Space Telescope mirror; it was less well-known for building the critical optics assemblies needed to produce the photo masks used in chip-making.

7. Gould Electronics，《一家有着深厚根基的年轻公司》，未注明日期的宣传册，约 1984 年；John Michels，《超级迷你汽车接替 T 型车》，Datamation（1974 年 2 月）：71-74 页。

7. Gould Electronics, “A Young Company with Deep Roots,” undated brochure, ca. 1984; John Michels, “The Mega-mini Succeeds the Model T,” Datamation (February, 1974): 71–74.

8.后来，“虚拟”一词作为“虚拟现实”一词的一部分而流行起来。它似乎起源于 IBM 推广 System=370 及其内存管理架构。这个词的使用可能来自文艺复兴时期艺术家的用法，他们谈到镜头或暗箱产生的“虚拟图像”。

8. The word “virtual” later became popular as part of the term “virtual reality.” It appears to have originated with IBM’s marketing of System=370 and its memory-management architecture. The use of that word may have come from its use among Rennaissance artists, who spoke of a “virtual image” produced by a lens or a camera obscura.

9. Arthur Burks、Herman Goldstine 和 John von Neumann，《电子计算仪器逻辑设计的初步讨论》，第二版，1947 年 9 月 2 日，（新泽西州普林斯顿：高等研究院）第 2 页，第 4-7 页；Simon Lavington，《曼彻斯特计算机的历史》，私人印刷，英国曼彻斯特，1975 年，第 32-33 页；T. Kilburn 等人，《单级存储系统》，IRE 电子计算机汇刊，EC-11（1962 年）：第 223-235 页。

9. Arthur Burks, Herman Goldstine, and John von Neumann, “Preliminary Discussion of the Logical Design of an Electronic Computing Instrument,” 2nd ed., 2 September 1947, (Princeton, NJ: Institute for Advanced Study) 2, 4–7; Simon Lavington, “History of Manchester Computers,” privately printed, Manchester, UK, 1975, 32–33; also T. Kilburn et al., “One-Level Storage System,” IRE Transactions on Electronic Computers, EC-11 (1962): 223–235.

10. Lavington，《曼彻斯特计算机的历史》，第 34 页。

10. Lavington, “History of Manchester Computers,” 34.

11. Franklin Fisher，《IBM 和美国数据处理行业》（纽约：Praeger，1983 年）：343-344。

11. Franklin Fisher, IBM and the US Data Processing Industry (New York: Praeger, 1983): 343–344.

12. Bell，Mudge 和 McNamara，《计算机工程：DEC 对硬件系统设计的看法》 (马萨诸塞州贝德福德：Digital Press，1978)：405-428。

12. Bell, Mudge, and McNamara, Computer Engineering: a DEC View of Hardware Systems Design (Bedford, MA: Digital Press, 1978): 405–428.

13.培生，《数字化工作》，73页。

13. Pearson, Digital at Work, 73.

14. Tracy Kidder，《新机器的灵魂》（波士顿：Little, Brown，1981 年）。Kidder 讲述了 Data General 如何拒绝使用 VAX 风格的“模式位”来提供与其旧产品线的兼容性。书中最戏剧性的情节之一描述了负责新计算机的工程师 Tom West 如何偷偷在客户现场打开 VAX 并将其拆开以查看其设计方式（31–32）。

14. Tracy Kidder, The Soul of a New Machine (Boston: Little, Brown, 1981). Kidder recounts how Data General resisted the use of a VAX-style “mode bit” to provide compatibility with its older line. One of the book’s most dramatic episodes describes how Tom West, the engineer in charge of the new computer, surreptitiously opened a VAX at a customer site and disassembled it to see how it was designed (31–32).

15.我无法证实这一说法，但从多个来源听说了这一说法。鉴于 DEC 对 UNIX 的支持力度很弱，这表明 Olsen 并不关心该操作系统；但其他一些更同情他的人说，他只是在指出一种普遍趋势（参见 JAWS），即每个人都想要 UNIX，尽管他们不知道该怎么使用它。

15. I have been unable to verify this statement but have heard it from several sources. In light of DEC’s weak support for UNIX, it suggests that Olsen did not care for the operating system; but others, more sympathetic, have said that he was only referring to a general trend (c.f. JAWS) that everyone wanted UNIX even though they did not know what to do with it.

16. 1988 年爆发的“互联网蠕虫”就是一个例证，它导致互联网瘫痪。该蠕虫是由康奈尔大学的一名学生编写的，利用了 VAX 系统软件中一些不为人知的缺陷。几年后，这种攻击造成的破坏性会小一些，因为 VAX 不再是主流机器。

16. One indication of this was the “Internet Worm,” unleashed in 1988, which brought the Internet down. It was written by a student at Cornell and took advantage of some obscure flaws in VAX system software. A few years later such an attack would have been less damaging because the VAX no longer was the dominant machine.

17. Fisher 等人，IBM 和美国数据处理行业，442--444。

17. Fisher et al., IBM and the U.S. Data Processing Industry, 442–-444.

18. Bob O. Evans，《IBM System=360》，《计算机博物馆报告》（1984 年夏季）：第 17 页。

18. Bob O. Evans, “IBM System=360,” Computer Museum Report (Summer 1984): 17.

19. DC Dykstra，《IBM 的信天翁：微型制造商的历史教训》，《Computerworld》第 18 卷（1984 年 12 月 10 日）：第 134 页。

19. D. C. Dykstra, “IBM’s Albatross: A History Lesson for Micro Makers,” Computerworld 18 (December 10, 1984): 134.

20.部分副本现存于特拉华州威尔明顿的哈格利博物馆和明尼苏达州明尼阿波利斯的查尔斯·巴贝奇研究所。以下审判摘要基于对哈格利记录的审查。与审判结果一致的审判概要可见于上述 Franklin Fisher 的两本书：Franklin Fisher 等人的《IBM和美国数据处理行业》（纽约：Praeger，1983 年）；Franklin Fisher、John J. McGowan 和 Joel E. Greenwood 的《折叠、纺锤和残缺：经济分析和美国诉 IBM》（剑桥：麻省理工学院出版社，1983 年）。Thomas Delamarter 的《蓝色巨人：IBM 的权力使用和滥用》（纽约：Dodd，Mead，1986 年）一书得出了相反的结论。

20. Partial copies are located at the Hagley Museum, Wilmington, Delaware, and at the Charles Babbage Institute, Minneapolis, Minnesota. The following summary of the trial is based on an examination of the transcripts at the Hagley. A synopsis of the trial, in agreement with its outcome, is found in Franklin Fisher’s two books, cited above: Franklin Fisher et al., the IBM and U.S. Data Processing Industry (New York: Praeger, 1983); and Franklin Fisher, John J. McGowan, and Joel E. Greenwood, Folded, Spindled, and Mutilated: Economic Analysis and U.S. vs. IBM (Cambridge: MIT Press, 1983). A book that draws the opposite conclusion is Thomas Delamarter, Big Blue: IBM’s Use and Abuse of Power (New York: Dodd, Mead, 1986).

21. 美国诉 IBM，F. Withington 的证词，55989。

21. U.S. v. IBM, testimony of F. Withington, 55989.

22.DeLamarter，《Big Blue》，xv。

22. DeLamarter, Big Blue, xv.

23. Paul Carroll，《Big Blues：IBM 的衰落》（纽约：Crown，1994 年）；Carroll 的书中所描述的 IBM 显然是自 1981 年个人计算机的推出开始的；另请参阅 Charles Ferguson 和 Charles Morris 的《计算机战争：IBM 的衰落和全球技术的未来》（纽约：Times Books，1994 年）。

23. Paul Carroll, Big Blues: the Unmaking of IBM (New York: Crown, 1994); the IBM that Carroll’s book describes is one that apparently began with the introduction of the personal computer in 1981; also Charles Ferguson and Charles Morris, Computer Wars: the Fall of IBM and the Future of Global Technology (New York: Times Books, 1994).

24.帕列夫斯基在不到十年的时间里积累了数亿美元的个人财富，法官对此很感兴趣。

24. The discussion of Palevsky’s amassing a personal fortune of hundreds of millions of dollars in less than a decade was notestext1d with some interest by the judge.

25. Fisher 等，IBM，438；另请参阅 Roy A. Bauer、Emilio Collar 和 Victor Tang 合著的《Silverlake 项目：IBM 的转型》（纽约：牛津大学出版社，1992 年）。

25. Fisher et al., IBM, 438; also Roy A. Bauer, Emilio Collar, and Victor Tang, The Silverlake Project: Transformation at IBM (New York: Oxford University Press, 1992).

26. Ivan T. Frisch 和 Howard Frank，“计算机通信：我们如何走到今天”，NCC 论文集44（1975）：109-117。

26. Ivan T. Frisch and Howard Frank, “Computer Communications: How We Got Where We Are,” Proceedings NCC 44 (1975): 109–117.

27. Lamont Wood，《发明个人电脑的人》，《美国发明与技术遗产》（1994 年秋季）：64；另见 Pearson 的《数字化工作》，第 90-92 页。

27. Lamont Wood, “The Man Who Invented the PC,” American Heritage of Invention & Technology (Fall 1994): 64; also Pearson, Digital At Work, 90–92.

28. Pugh，IBM 的 360，606。

28. Pugh, IBM’s 360, 606.

29.同上，第545–549页。

29. Ibid., 545–549.

30.同上，550。

30. Ibid., 550.

31. “AESOP：一种通用的实时直接访问管理信息系统方法” (马萨诸塞州贝德福德：MITRE 公司，1966 年 6 月)，报告 AD-634371，第 7 页。

31. “AESOP: A General Purpose Approach to Real-Time, Direct Access Management Information Systems” (Bedford, MA: MITRE Corporation, June 1966), Report AD-634371, 7.

32. Datamation (1968 年 10 月)：17；另请参阅 Robert Glass 的《计算灾难》 (西雅图：计算趋势，1983 年)：57-69；另请参阅 W. David Gardner 的《128 号公路：波士顿的技术温床》，Datamation (1981 年 11 月)：110-118。

32. Datamation (October 1968): 17; also Robert Glass, Computing Catastrophes (Seattle: Computing Trends, 1983): 57–69; also W. David Gardner, “Route 128: Boston’s Hotbed of Technology,” Datamation (November 1981): 110–118.

33.同上；另见 Viatron 文件，盒子 A30，波士顿计算机博物馆，历史收藏品。

33. Ibid.; also Viatron file, Box A30, Computer Museum, Boston, Historical Collections.

34.丹尼尔·惠特尼致计算机博物馆的信，同上。

34. Letter, Daniel Whitney to Computer Museum, Ibid.

35. An Wang，《Lessons》（马萨诸塞州雷丁：Addison-Wesley，1986 年）。

35. An Wang, Lessons (Reading, MA: Addison-Wesley, 1986).

36. Pearson，《数字化工作》，38；CE MacKenzie，《编码字符集：历史与发展》（马萨诸塞州雷丁：Addison-Wesley，1980 年）；Calvin Mooers，与美国国家历史博物馆史密森尼计算机历史项目的 Jon Eklund 的访谈。

36. Pearson, Digital at Work, 38; C. E. MacKenzie, Coded Character Sets: History & Development (Reading, MA: Addison-Wesley, 1980); Calvin Mooers, interview with Jon Eklund, Smithsonian Computer History Project, National Museum of American History.

37.Pugh，IBM 的 360，613。

37. Pugh, IBM’s 360, 613.

38. Edwin McDowell，《“没有问题”的机器给总统带来了问题》，《纽约时报》（1981 年 3 月 24 日）：C-7；另见同上，3 月 20 日、26 日、3 月 16 日、1 日和 3 月 27 日、26 日。《纽约时报》3月 20 日的社论讽刺性地哀叹，文字处理器将剥夺未来历史学家发掘伟大人物早期思想的乐趣，这些思想都记录在手稿草稿中。

38. Edwin McDowell, “‘No Problem’ Machine Poses a Presidential Problem,” New York Times (March 24, 1981): C-7; see also Ibid., March 20, 26; March 16, 1, and March 27, 26. The Times editorial on March 20, with tongue in cheek, lamented that word processors would deprive future historians of the joy of uncovering a great figure’s early thoughts, as recorded on rough drafts of manuscripts.

39. Charles Kenney，《骑着脱缰的野马：王安实验室的崛起与衰落》（波士顿：Little, Brown，1992 年）。

39. Charles Kenney, Riding the Runaway Horse: the Rise and Decline of Wang Laboratories (Boston: Little, Brown, 1992).

40. Kenney，《骑着脱缰的马》，68–73；另请参阅 Wang，《教训》，182。

40. Kenney, Riding the Runaway Horse, 68–73; also Wang, Lessons, 182.

41. Datamation (1976 年 6 月)：48-61；另见 1985 年 6 月 1 日第 50-51、65 页；另见 Stephen T. McClellan 所著《即将到来的计算机行业大萧条：赢家、输家和幸存者》(纽约：Wiley，1984 年)，第 15 章。

41. Datamation (June 1976): 48–61; also June 1, 1985, 50–51, 65; also Stephen T. McClellan, The Coming Computer Industry Shakeout: Winners, Losers, and Survivors (New York: Wiley, 1984), chapter 15.

42.下面的叙述基于一些二手资料，主要是道格拉斯·史密斯 (Douglas Smith) 和罗伯特·亚历山大 (Robert Alexander )的《摸索未来：施乐公司如何发明，然后又忽略了第一台个人电脑》 (纽约：William Morrow，1988 年) 和乔治·帕克 (George Pake) 的《施乐 PARC 的研究：一位创始人的评估》(Research at Xerox PARC: a Founder's Assessment)，《IEEE Spectrum》(1985 年 10 月)：54-75。

42. The following account is based on a number of secondary sources, primarily Douglas Smith and Robert Alexander, Fumbling the Future: How Xerox Invented, Then Ignored, the First Personal Computer (New York: William Morrow, 1988), and George Pake, “Research at Xerox PARC: a Founder’s Assessment,” IEEE Spectrum (October 1985): 54–75.

43.转引自戴维·迪克森著《科学的新政治》（纽约：万神殿图书，1984 年）：第 122 页。

43. Quoted in David Dickson, The New Politics of Science (New York: Pantheon Books, 1984): 122.

44. Arthur Norberg、Judy O'Neill 与 Kerry Freedman 合著，《国防高级研究计划局信息处理技术办公室史》（明尼苏达州明尼阿波利斯：查尔斯·巴贝奇研究所，1992 年）。

44. Arthur Norberg and Judy O’Neill, with Kerry Freedman, “A History of the Information Processing Techniques Office of the Defense Advanced Research Projects Agency” (Minneapolis, MN: Charles Babbage Institute, 1992).

45.同上；另请参阅 C. Gordon Bell 和 John E. McNamara 合著的《高科技企业：创业成功指南》（马萨诸塞州雷丁：Addison-Wesley，1991 年）：第 101 页；另请参阅 Pake 的《施乐 PARC 的研究》。Metcalfe 正在哈佛大学攻读博士学位，但被 PARC 招募时，他在麻省理工学院获得了 ARPA 资助的工作。

45. Ibid.; also C. Gordon Bell and John E. McNamara, High Tech Ventures: the Guide for Entrepreneurial Success (Reading, MA: Addison-Wesley, 1991): 101; also Pake, “Research at Xerox PARC.” Metcalfe was getting his Ph.D. from Harvard, but at the time he was recruited by PARC he had an ARPA-funded job at MIT.

46. JCR Licklider，“人机共生”，IRE Transactions on Human Factors 1（1960 年 3 月）：4-11；Licklider 和 Taylor，“计算机作为一种通讯设备”，Science and Technology（1968 年 4 月）。

46. J. C. R. Licklider, “Man-Computer Symbiosis,” IRE Transactions on Human Factors 1 (March 1960): 4–11; Licklider and Taylor, “The Computer as a Communication Device,” Science and Technology (April 1968).

47. Norberg 和 O'Neill，“信息处理技术史”，33-60。

47. Norberg and O’Neill, “A History of the Information Processing Techniques,” 33–60.

48. Engelbart，Adele Goldberg 编，《个人工作站的历史》（马萨诸塞州雷丁：Addison-Wesley，1988 年）：191。

48. Engelbart, in Adele Goldberg, ed., A History of Personal Workstations (Reading, MA: Addison-Wesley, 1988): 191.

49. William English、Douglas Engelbart 和 Melvyn Berman，“文本操作的显示选择技术”，IEEE 电子学人为因素学报8（1967 年 3 月）：5-15。

49. William English, Douglas Engelbart, and Melvyn Berman, “Display-Selection Techniques for Text Manipulation,” IEEE Transactions on Human Factors in Electronics 8 (March 1967): 5–15.

50. Douglas C. Engelbart 和 William English，《增强人类智力的研究中心》，《秋季 JCC 论文集》 33-1（1968 年）：395-410 页；另请参阅 Goldberg，《个人工作站的历史》，第 202-206 页。

50. Douglas C. Engelbart and William English, “A Research Center for Augmenting Human Intellect,” Proceedings Fall JCC 33-1 (1968): 395–410; also Goldberg, History of Personal Workstations, 202–206.

51. Douglas Smith 和 Robert Alexander，《摸索未来：施乐公司如何发明，然后又忽视了第一台个人电脑》（纽约：William Morrow，1988 年）；Robert Metcalfe，《以太网是如何发明的》，《计算机历史年鉴》 16：4（1994 年）：81-88；Tekla Perry 和 John Voelcker，《鼠标和菜单：设计用户友好界面》，《IEEE Spectrum》（1989 年 9 月）：46-51。

51. Douglas Smith and Robert Alexander, Fumbling the Future: How Xerox Invented, Then Ignored, the First Personal Computer (New York: William Morrow, 1988); Robert Metcalfe, “How Ethernet was Invented,” Annals of the History of Computing 16: 4 (1994): 81–88; Tekla Perry and John Voelcker, “Of Mice and Menus: Designing the User-Friendly Interface,” IEEE Spectrum (September 1989): 46–51.

52. Larry Press，《Altair 之前：个人计算机的历史》(1993 年)：第 27-33 页。

52. Larry Press, “Before the Altair: the History of the Personal Computer,” (1993): 27–33.

53. Goldberg，《个人工作站史》，第 265–289 页。显然，Flip Wilson 在 1969 年的一集节目中即兴发挥了这句短语，当时他扮成了他的另一个自我 Geraldine Jones；参见《计算机史年鉴》第 17 卷：1 期（1995 年），第 5 页。

53. Goldberg, History of Personal Workstations, 265–289. Apparently Flip Wilson ad-libbed the phrase on an episode in 1969, while cross-dressed as his alter ego Geraldine Jones; see Annals of the History of Computing 17: 1 (1995), 5.

54. David Smith 等人，“设计星型用户界面”，Byte（1982 年 4 月）：242–282。

54. David Smith et al., “Designing the Star User Interface,” Byte (April 1982): 242–282.

55. Phillip Ein-Dor，“再论Grosch定律”，CACM 28:2（1985）：142-151。

55. Phillip Ein-Dor, “Grosch’s Law Re-revisited,” CACM 28: 2 (1985): 142–151.

56. Peggy Kidwell 和 Paul Ceruzzi，《数字计算的里程碑》（华盛顿特区：史密森学会出版社，1994 年）：97。

56. Peggy Kidwell and Paul Ceruzzi, Landmarks in Digital Computing (Washington, DC: Smithsonian Institution Press, 1994): 97.

57.史蒂文·曼内斯和保罗·安德鲁斯，《盖茨：微软大亨如何重塑一个行业并成为美国首富》（纽约：Doubleday，1993 年）：第 111 页。在撰写本文时（1997 年），微软已同意向苹果公司投资数亿美元以拯救它。

57. Steven Manes and Paul Andrews, Gates: How Microsoft’s Mogul Reinvented an Industry, and Made Himself the Richest Man in America (New York: Doubleday, 1993): 111. As this is being written (1997), Microsoft has agreed to invest a few hundred million dollars in Apple to rescue it.

58. Steven Wozniak，《苹果 II》，Byte（1977 年 5 月）；Gregg Williams 和 Rob Moore 对 Wozniak 的采访，《苹果的故事，第 2 部分》，Byte（1985 年 1 月）：167–180。

58. Steven Wozniak, “The Apple II,” Byte (May 1977); also interview of Wozniak by Gregg Williams and Rob Moore, “The Apple Story, Part 2,” Byte (January 1985): 167–180.

59. Steven Wozniak，《工程师和计算机的诞生》，《计算机博物馆报告》（1986 年秋季）：3-8；Gregg Williams 和 Rob Moore 也对其进行了采访，《Byte》（1985 年 1 月）：167-172。

59. Steven Wozniak, “The Making of an Engineer and a Computer,” Computer Museum Report (Fall 1986): 3–8; also interview by Gregg Williams and Rob Moore, Byte (January 1985): 167–172.

60.苹果广告，Byte（1978年7月）：14-15。

60. Advertisement for Apple, Byte (July 1978): 14–15.

61. Steven Burke，《Visicalc 说再见》，《信息世界》（1985 年 6 月 24 日）：20-21；Daniel Bricklin，《Visicalc 与软件艺术：从创世纪到出埃及记》，《计算机博物馆报告》（1986 年夏）：8-10；Susan Lammers 主编，《程序员在工作》（华盛顿州雷德蒙德：微软出版社，1989 年）：131-160。

61. Steven Burke, “Visicalc Says Goodbye,” Infoworld (June 24, 1985): 20–21; also Daniel Bricklin, “Visicalc and Software Arts: Genesis to Exodus,” Computer Museum Report (Summer 1986): 8–10; also Susan Lammers, ed., Programmers at Work (Redmond, WA: Microsoft Press, 1989): 131–160.

62. Briklin，《计算机博物馆报告》，同上，第 9 页。

62. Briklin, in Computer Museum Report, ibid., 9.

63. IBM PC 本身并没有限制只能寻址 640 K 内存，但很快这成为了事实上的限制。它很快就成为 PC 系列计算机的祸根。

63. The IBM PC was not inherently restricted to addressing only 640 K of memory, but soon that became a de facto limit. It soon became the curse of the PC line of computers.

64. Jan Chposky，《蓝色魔法：IBM 个人计算机背后的人、权力和政治》（纽约：Facts on File，1988 年）；另见《年度机器：计算机的兴起》，《时代》（1983 年 1 月 3 日）：封面，第 14-37 页。

64. Jan Chposky, Blue Magic: the People, Power and Politics Behind the IBM Personal Computer (New York: Facts on File, 1988); also “Machine of the Year: the Computer Moves In,” Time (January 3, 1983): cover, 14–37.

65. David Bradley，《IBM PC 的诞生》，《Byte》（1990 年 9 月）：414–420。

65. David Bradley, “The Creation of the IBM PC,” Byte (September 1990): 414– 420.

66.这个故事有很多版本，包括谁选择了 8088 芯片。在这个简短的总结中，我参考了盖茨第 11 章中 Manes 和 Andrews 的叙述。

66. There are many variations of this story, including who chose the 8088 chip. In this brief summary I have relied on the account of Manes and Andrews in Gates, chapter 11.

67.许多人指责 IBM 盗用了小公司的技术，却没有给予其创造者认为公平的补偿。例如，参见 An Wang 对其磁芯存储器专利的指控，载于《Lessons》（马萨诸塞州雷丁：Addison-Wesley，1986 年）；以及 Erwin Tomash 和 Arnold Cohen 对 ERA 磁鼓存储器开发的描述，《ERA 的诞生：工程研究协会，1946-1955 年》，《计算机历史年鉴》 1（1979 年）：83-97。

67. There have been many charges that IBM appropriated the technology of small companies without giving what their creators felt was fair compensation. See, for example, An Wang’s charge regarding his patent on core memory, in Lessons (Reading, MA: Addison-Wesley, 1986); and Erwin Tomash and Arnold Cohen’s account of ERA’s development of the drum memory, “The Birth of an ERA: Engineering Research Associates, Inc., 1946–1955,” Annals of the History of Computing 1 (1979): 83–97.

68. Manes 和 Andrews，《Gates》，160；还有 Tim Paterson 与作者的电话采访，1996 年 7 月 23 日。

68. Manes and Andrews, Gates, 160; also Tim Paterson, telephone interview with the author, 23 July 1996.

69.采访帕特森。1996 年 PBS 电视连续剧《书呆子的胜利》强烈暗示 MD-DOS 是从 CP=M“窃取”的，但没有提供任何证据。另请参阅 G. Pascal Zachary 的《致命缺陷》，Upside（1994 年 11 月）：18-27。

69. Interview with Paterson. A 1996 PBS television series, “Triumph of the Nerds,” strongly insinuated that MD-DOS was “stolen” from CP=M, without offering any proof. See also G. Pascal Zachary, “Fatal Flaw,” Upside (November 1994): 18–27.

70.例如，在 CP=M 中，命令 PIP A:*.* B:*.* 将第二个磁盘驱动器上的所有文件复制到第一个驱动器。要在 MS-DOS 中执行此操作，请输入 COPY B: *.* A:。

70. For example, in CP=M the command PIP A:*.* B:*.* copied all the files on the second disk drive over to the first drive. To do that with MS-DOS one would write COPY B: *.* A:.

71.上述观察主要基于作者所掌握的 CP=M 和 MS-DOS 手册。

71. The above observations are based primarily on CP=M and MS-DOS manuals in the author’s possession.

72.佩吉·瓦特（Peggy Watt）转述自《信息世界》（1991 年 8 月 12 日）：第 48 页。

72. Quoted by Peggy Watt in Infoworld (Aug. 12, 1991): 48.

73.蒂姆·帕特森，1996 年 7 月 23 日与作者进行电话采访。

73. Tim Paterson, telephone interview with the author, 23 July 1996.

74. Bradley，《IBM PC 的诞生》，第 420 页。

74. Bradley, “The Creation of the IBM PC,” 420.

75. Chposky，《蓝色魔法》，180。

75. Chposky, Blue Magic, 180.

76. George Basalla，《技术的进化》（英国剑桥：剑桥大学出版社，1988 年）。

76. George Basalla, The Evolution of Technology (Cambridge, UK: Cambridge University Press, 1988).

77. Jef Raskin，致编辑的信，Datamation（1976 年 8 月）：第 8 页；另请参阅 Raskin 的《历史中的漏洞：为何苹果年鉴不可靠》，手稿私下流传，1994 年。

77. Jef Raskin, letter to the editor, Datamation (August 1976): 8; also Raskin, “Holes in the Histories: Why the Annals of Apple have been Unreliable,” MS privately circulated, 1994.

78. Raskin 采访了 Susan Lammers 主编的《程序员在工作》（华盛顿州雷德蒙德：微软出版社，1989 年）：227-245；此外还采访了 Ronald Baecker 和 William AS Buxton 的《人机交互读物：一种多学科方法》（加利福尼亚州洛斯阿尔托斯：Morgan Kaufmann，1987 年）：649-667。

78. Raskin, interviewing Susan Lammers, ed., Programmers at Work (Redmond, WA: Microsoft Press, 1989): 227–245; also Ronald Baecker and William A. S. Buxton, Readings in Human-Computer Interaction: a Multidisciplinary Approach (Los Altos, CA: Morgan Kaufmann, 1987): 649–667.

79.例如，参见 Steven Levy 的《Insanely Great》（纽约：Viking，1994 年）。Levy 的书有许多事实错误，因此不值得信赖；最可靠的记载是 Fred Guterl 的《设计案例历史：苹果的 Macintosh》，IEEE Spectrum（1984 年 12 月）：34-43。

79. See, for example, Steven Levy, Insanely Great (New York: Viking, 1994). Levy’s book has so many factual errors that it cannot be relied upon, however; the most reliable account is Fred Guterl, “Design Case History: Apple’s Macintosh,” IEEE Spectrum (December 1984): 34–43.

80. Raskin，载于Lammers著《程序员在工作》，第230页。

80. Raskin, in Lammers, Programmers at Work, 230.

81.确实，没有特殊工具是无法打开机箱的。不久之后，第三方供应商开始销售“Mac Cracker”，它结合了打开机箱所需的特殊“torx”螺丝刀和撬具。

81. Indeed, one could not open the case without special tools. It was not long before third-party vendors began selling a “Mac Cracker” that combined the special “torx” screwdriver and prying tool need to open the case.

82. Tom Thompson，《Macintosh 10 周年》，Byte（1994 年 2 月）：47–54。

82. Tom Thompson, “The Macintosh at 10,” Byte (February 1994): 47–54.

83.Datamation (1985年6月1 日):139-140。

83. Datamation (June 1, 1985): 139–140.

84.本文基于对当时Infoworld期刊的扫描。据说，需要运行 Lotus 1-2-3 作为兼容性测试是 640 K 内存壁垒变得如此根深蒂固的主要原因。1-2-3 将 640 K 以上的内存地址用于其他功能，从而阻止该段用于一般存储。

84. This passage is based on a scanning of the issues of Infoworld during that period. The need to run Lotus 1-2-3 as a test of compatibility is said to have been the main reason that the 640 K memory barrier became so entrenched. 1-2-3 used memory addresses above 640 K for other functions, thus precluding that segment from ever being used for general storage.

第九章

Chapter 9

1. “分布式操作多路访问交互网络。”

1. “Distributive Operating Multi-Access Interactive Network.”

2. Mark Hall 和 John Barry，《Sunburst：Sun Microsystems 的崛起》（芝加哥：当代图书出版社，1990 年）：第 60-61 页；C. Gordon Bell 和 John E. McNamara，《高科技企业：创业成功指南》（马萨诸塞州雷丁：Addison-Wesley 出版社，1991 年）：第 39-42 页、第 323-325 页。

2. Mark Hall and John Barry, Sunburst: the Ascent of Sun Microsystems (Chicago: Contemporary Books, 1990): 60–61; also C. Gordon Bell and John E. McNamara, High Tech Ventures: the Guide for Entrepreneurial Success (Reading, MA: Addison-Wesley, 1991): 39–42, 323–325.

3. Stephen T. McClellan，《即将到来的计算机行业大萧条：赢家、输家和幸存者》（纽约：Wiley，1984 年）：280-281。

3. Stephen T. McClellan, The Coming Computer Industry Shakeout: Winners, Losers, & Survivors (New York: Wiley, 1984): 280–281.

4. Hall 和 Barry，《Sunburst》，第 1 章；Bell 和 McNamara，《High-Tech Ventures》，325–326。

4. Hall and Barry, Sunburst, chapter 1; Bell and McNamara, High-Tech Ventures, 325–326.

5. Peter Salus，《UNIX 的四分之一世纪》（马萨诸塞州雷丁：Addison-Wesley，1994 年）；DM Ritchie 的《Unix 分时系统：回顾》，《贝尔系统技术期刊》第 57 期（1978 年）：1947-1969 年。

5. Peter Salus, A Quarter Century of UNIX (Reading, MA: Addison-Wesley, 1994); also D. M. Ritchie, “Unix Time-Sharing System: a Retrospective,” Bell System Technical Journal 57 (1978): 1947–1969.

6. Salus，《UNIX 的四分之一世纪》，137–145。

6. Salus, A Quarter Century of UNIX, 137–145.

7.另请参阅“ed”、“ln”、“mv”等。

7. See also “ed,” “ln,” “mv,” and many others.

8. Donald A. Norman，《UNIX 的麻烦》，Datamation（1981 年 11 月）：139-150。

8. Donald A. Norman, “The Trouble with UNIX,” Datamation (November 1981): 139–150.

9. Salus，《四分之一世纪》，137–142；153–172。

9. Salus, Quarter Century, 137–142; 153–172.

10.同上，第 153–172 页。其他记载与 Salus 的说法不同，它们声称 Bolt Beranek 和 Newman 根据 ARPA 合同负责 UNIX 中的 TCP=IP。

10. Ibid., 153–172. Other accounts differ with Salus and state that Bolt Beranek and Newman, under an ARPA contract, was responsible for TCP=IP in UNIX.

11. Jamie Pearson，《数字化工作》（马萨诸塞州贝德福德：Digital Press，1992 年）：70-73；还有 C. Gordon Bell 和 John E. McNamara 合著的《高科技企业：创业成功指南》（马萨诸塞州雷丁：Addison-Wesley，1991 年）：37。

11. Jamie Pearson, Digital at Work (Bedford, MA: Digital Press, 1992): 70–73; also C. Gordon Bell and John E. McNamara, High Tech Ventures: the Guide for Entrepreneurial Success (Reading, MA: Addison-Wesley, 1991): 37.

12. Glenn Rifkin 和 George Harrar，《终极企业家：肯·奥尔森和数字设备公司的故事》（芝加哥：当代图书，1988 年），第 25、29、30 章。Rainbow 设计精良，几乎可以与 IBM 兼容。但“几乎”还不够好——这一事实在开始时只有少数人意识到，但到了 1982 年，康柏等公司认识到这是与 IBM 竞争的唯一途径。

12. Glenn Rifkin and George Harrar, The Ultimate Entrepreneur: the Story of Ken Olsen and Digital Equipment Corporation (Chicago: Contemporary Books, 1988), chapters 25, 29, 30. The Rainbow was well-engineered and almost IBM compa tible. But almost was not good enough–-a fact that only a few realized at the outset, but which by 1982 was recognized by companies like Compaq as the only way to compete against IBM.

13. C. Gordon Bell、J. Craig Mudge 和 John McNamara，《计算机工程：DEC 对硬件系统设计的视角》 (马萨诸塞州贝德福德：Digital Press，1978 年)，第 17 章。

13. C. Gordon Bell, J. Craig Mudge, and John McNamara, Computer Engineering: a DEC View of Hardware Systems Design (Bedford, MA: Digital Press, 1978), chapter 17.

14. David A. Patterson，“精简指令集计算机”，CACM 28（1985）：8-21。

14. David A. Patterson, “Reduced Instruction Set Computers,” CACM 28 (1985): 8–21.

15. John Markoff，《特立独行的科学家获得 IBM 致敬》，《纽约时报》，1990 年 6 月 26 日，D1；Herbert Grosch 在《计算机：生活中的片段》（加利福尼亚州诺瓦托：Third Millenium Books，1991 年）：258 中描述了“野鸭”备忘录。

15. John Markoff, “A Maverick Scientist Gets an I.B.M. Tribute,” New York Times, 26 June 1990, D1; the “wild duck” memo is described by Herbert Grosch in Computer: Bit Slices from a Life (Novato, CA: Third Millenium Books, 1991): 258.

16. George Radin，《801 小型计算机》，IBM J. Res. Dev . 27（1983 年 5 月）：237–246。

16. George Radin, “The 801 Minicomputer,” IBM J. Res. Dev. 27 (May 1983): 237–246.

17. Patterson，《精简指令集计算机》，16、20；另请参阅 John L. Hennessy 和 David Patterson 合著的《计算机架构：一种定量方法》（加利福尼亚州圣马特奥：Morgan Kaufmann，1990 年）。

17. Patterson, “Reduced Instruction Set Computers,” 16, 20; also John L. Hennessy and David Patterson, Computer Architecture: a Quantitative Approach (San Mateo, CA: Morgan Kaufmann, 1990).

18. R. Emmett Carlyle，“RISC-Y 业务？” Datamation（1985 年 2 月 15 日）：30-35。

18. R. Emmett Carlyle, “RISC-Y Business?” Datamation (February 15, 1985): 30– 35.

19. John Hennessy 和 Norman Jouppi，“计算机技术与架构：不断发展的互动”，IEEE 计算机24（1991）：18-29。

19. John Hennessy and Norman Jouppi, “Computer Technology and Architecture: an Evolving Interaction,” IEEE Computer 24 (1991): 18–29.

20. Hennessy 和 Patterson，《计算机架构》，190；Hall 和 Barry，《Sunburst》，163。

20. Hennessy and Patterson, Computer Architecture, 190; Hall and Barry, Sunburst, 163.

21. Hennessy 和 Patterson，《计算机架构》；Silicon Graphics Inc.，1989–1993 年年度报告。

21. Hennessy and Patterson, Computer Architecture; Silicon Graphics Inc., Annual Reports for 1989–1993.

22.Grosch，计算机，130-131。

22. Grosch, Computer, 130–131.

23. Robert M. Metcalfe，《以太网是如何发明的》，《计算机史年鉴》 16：（1994）：81-88。

23. Robert M. Metcalfe, “How Ethernet was Invented,” Annals of the History of Computing 16: (1994): 81–88.

24. Metcalfe，“以太网是如何发明的”，83页。

24. Metcalfe, “How Ethernet was Invented,” 83.

25. R. Binder、N. Abramson、F. Kuo、A. Okinaka 和 D. Wax，《ALOHA 数据包广播：回顾》，载于 Siewiorek 等人的《计算机结构》，第 416-428 页。

25. R. Binder, N. Abramson, F. Kuo, A. Okinaka, and D. Wax, “ALOHA Packet Broadcasting: a Retrospect,” in Siewiorek et al., Computer Structures, 416–428.

26. “以太”一词源于“发光以太”，物理学家认为它能携带光，至少在迈克尔逊和莫利于 1887 年进行著名实验而无法找到其存在的证据之前是如此。物理学家不再相信以太的存在，但计算机科学家却对其了如指掌。

26. The term “ether” came from the “luminiferous aether” that physicists believed carried light, at least until Michaelson and Morley were unable to find evidence of its existence in their famous experiment of 1887. Physicists no longer believe in the existence of the ether, but computer scientists know it well.

27. Robert M. Metcalfe 和 David R. Boggs，《以太网：本地计算机网络的分布式分组交换》，Siewiorek 等著，《计算机结构》，429-438。

27. Robert M. Metcalfe, and David R. Boggs, “Ethernet: Distributed Packet Switching for Local Computer Networks,” in Siewiorek et al., Computer Structures, 429–438.

28. Metcalfe，“以太网是如何发明的”，85。

28. Metcalfe, “How Ethernet was Invented,” 85.

29. C. Gordon Bell，收录于 Adele Goldberg 编的《个人工作站的历史》（纽约：ACM Press，1988 年）：第 19 页。IBM 网络是一个令牌环系统，在这种拓扑结构中，对通道的访问由持有“令牌”的计算机控制，就像在美国铁路发展的早期，工程师必须持有唯一的令牌，才能在没有信号的轨道上驾驶火车，以防止发生碰撞。

29. C. Gordon Bell, in Adele Goldberg, ed., A History of Personal Workstations (New York: ACM Press, 1988): 19. The IBM network was a Token Ring system, a topology in which access to the channel was controlled by whichever computer held a “token,” just as in the early days of U.S. railroads an engineer had to hold a unique token before he was allowed to take a train on a piece of unsignaled track, to prevent collisions.

30.华尔街是一个例外，精通计算机的股票分析师在 SUN 工作站上开发了复杂的程序来跟踪价格变动并建议何时买入或卖出股票。

30. One exception was Wall Street, where computer-savvy stock analysts developed sophisticated programs on SUN workstations to track price movements and recommend when to buy or sell a stock.

31.Byte （1984年12月）： 148。

31. Byte (December 1984): 148.

32.截至本文撰写之时，这一等式似乎可能会发生变化。Linux 是一款免费的 UNIX 系统，可在先进的 Intel 微处理器上运行良好，它为个人计算机所有者提供了上一代工作站的大部分功能。同样，微软的操作系统“Windows NT”是基于 UNIX 的工作站的直接竞争对手，也可以在个人计算机上运行。也许工作站可能会因此而消失。

32. As of this writing it appears that this equation may change. Linux, a free UNIX system that runs well on advanced Intel microprocessors, gives the owners of personal computers most of the power of an earlier generation of workstations. Likewise, Microsoft’s operating system “Windows NT” is a direct competitor to UNIX-based workstations and also runs on personal computers. Perhaps the workstation may vanish as a class as a result.

33. “分组交换”一词可能起源于英国国家物理实验室的 Donald Davies。参见 Martin Campbell-Kelly，《国家物理实验室的数据通信（1965-1975）》，《计算机历史年鉴》第 9 卷（1988 年）：221-247 页。兰德公司的 Paul Baran 可能同时独立发现了该术语。兰德公司的这项工作最初是保密的。

33. The term “Packet Switching” probably originated with Donald Davies of the National Physical Laboratory in the U.K. See Martin Campbell-Kelly, “Data Communications at the National Physical Laboratory (1965–1975),” Annals of the History of Computing 9 (1988): 221–247. It may have been independently discovered by Paul Baran of the RAND Corporation at the same time. The RAND work was initially classified.

34. Janet Abbate，“从 ARPANET 到互联网：ARPA 赞助的计算机网络史，1966-1988”(博士论文，宾夕法尼亚大学，1994 年)：第 109 页。

34. Janet Abbate, “From ARPANET to Internet: a History of ARPA-Sponsored Computer Networks, 1966–1988” (Ph.D. diss., University of Pennsylvania, 1994): 109.

35.本节我参考了 Judy O'Neill、Janet Abbate 和 Juan Rogers 尚未发表的著作。我非常感谢他们和其他与我分享初步成果的人。一些互联网的创造者撰写了《互联网简史》，不出所料，这本书只能在互联网上找到。他们在CACM（1997 年 2 月）上发表了一个缩略版。

35. For this section I am relying on the mainly unpublished work of Judy O’Neill, Janet Abbate, and Juan Rogers. I am grateful to them and to others who have shared their preliminary work with me. Some of the Internet’s creators have written “A Brief History of the Internet,” which, not surprisingly, is available only on the Internet itself. They have published an abbreviated version in CACM (February 1997).

36. ARPANET 最初使用不同的协议 NCP 建立，但后来发现它不适合将不同的网络连接在一起。ARPANET 本身于 1983 年 1 月转向 TCP=IP。

36. ARPANET was initially set up using a different protocol, NCP, but it was found to be ill-suited to connecting different networks to one another. ARPANET itself shifted to TCP=IP in January 1983.

37. Bob Metcalfe，《应该有一项法律》，《纽约时报》，1996 年 7 月 15 日，C5。

37. Bob Metcalfe, “There Oughta be a Law,” New York Times, 15 July 1996, C5.

38. Peter H. Salus，《撒网：从 ARPANET 到 INTERNET 及其他》（马萨诸塞州雷丁：Addison-Wesley，1995 年）；第 5 章和第9章。

38. Peter H. Salus, Casting the Net: From ARPANET to INTERNET and Beyond (Reading, MA: Addison-Wesley, 1995); chapters 5 and 9.

39.大多数早期调制解调器的工作速度为 300 波特，这与每秒 300 位并不完全相同，但在同一范围内。

39. Most early modems worked at 300 Baud, which is not exactly the same as 300 bits per second but is in the same range.

40. Abbate，《从阿帕网到互联网》；另见 Ed Krol，《整个互联网用户指南和目录》（加利福尼亚州塞巴斯托波尔：O'Reilly & Associates，1992 年）：128-130。

40. Abbate, “From Arpanet to Internet”; also Ed Krol, The Whole Internet Users’ Guide and Catalog (Sebastopol, CA: O’Reilly & Associates, 1992): 128–130.

41.万尼瓦尔·布什，《正如我们所想》，《大西洋月刊》，1945 年。本文已重印多次；有关其写作、出版和早期影响的讨论，请参阅 James Nyce 和 Paul Kahn 编，《从 Memex 到超文本：万尼瓦尔·布什和心灵机器》（波士顿：Academic Press，1991 年）。

41. Vannevar Bush, “As We May Think,” Atlantic Monthly, 1945. The essay has been reprinted many times; for a discussion of its writing, publication, and early impact, see James Nyce and Paul Kahn, eds., From Memex to Hypertext: Vannevar Bush and the Mind’s Machine (Boston: Academic Press, 1991).

42. Nelson，《梦想机器》，DM 44，45。

42. Nelson, in Dream Machines, DM 44, 45.

43.同上，DM 19。

43. Ibid., DM 19.

44. Engelbart 的 NLS（在线系统）逐渐衰落，但后来出现了用于个人计算机的大纲程序，它们在个人计算机中占据了一席之地，虽然规模不大，但经久不衰。例如 Thinktank、Lotus Agenda 和 Ecco。

44. Engelbart’s NLS (On-Line System) faded, but outliner programs later appeared for personal computers, where they have established a small but persistent niche. Examples include Thinktank, Lotus Agenda, and Ecco.

45.超文本方面的学术工作在《ACM 通讯》 31的特刊（1988 年 7 月）中进行了总结。

45. Academic work in Hypertext was summarized in a special issue of the Communications of the ACM 31 (July 1988).

46.巧合的是，上文引用的CACM超文本特刊中一封致编辑的信正是由 ARPA 的两位项目经理写的，他们讨论了 ARPANET 即将解体以及网络活动转移到其他地方的问题。

46. By coincidence, one of the letters to the editor of the special issue of the CACM on hypertext, cited above, was by two program managers at ARPA, who discussed the impending dismantling of the ARPANET and the shifting of network activities elsewhere.

47. Tim Berners-Lee，《WWW：过去、现在和未来》，IEEE 计算机29（1996 年 10 月）：第 69-77 页。Berners-Lee 明确提到 Vannevar Bush、Doug Engelbart 和 Ted Nelson 是万维网概念的创始人。

47. Tim Berners-Lee, “WWW: Past, Present, and Future,” IEEE Computer 29 (October 1996): 69–77. Berners-Lee explicitly mentions Vannevar Bush, Doug Engelbart, and Ted Nelson as originators of the concepts that went into the Web.

48.同上，第70页。

48. Ibid., 70.

49.同上，第71页。

49. Ibid., 71.

50.安德森在互联网上的一些帖子已被保存在电子档案中。由于无法知道这些材料将保存多久，或者是否仍可供学者查阅，因此我没有在这里引用。

50. Some of Andreesen’s postings on the Internet have been preserved in an electronic archive. Since there is no way of knowing how long this material will be preserved, or whether it will remain accessible to scholars, I have not cited it here.

51.本年表主要取自《商业周刊》 1996年7月15日第56-59页的一篇文章。

51. This chronology is mainly taken from an article in Business Week, July 15, 1996, 56–59.

52. IBM 即使在大萧条时期也盈利，但 1991 年亏损 28 亿美元，1992 年亏损 50 亿美元，1993 年亏损 80 亿美元。1995 年，该公司恢复盈利。DEC 在 1994 年亏损约 20 亿美元，直到 1995 年中期才勉强恢复盈利。1998 年初，DEC 被卖给了康柏公司。

52. IBM made profits even during the Great Depression, but it lost $2.8 billion in 1991, $5 billion in 1992, and $8 billion in 1993. It returned to profitability in 1995. DEC lost around $2 billion in 1994 and just barely started making money again in mid-1995. In early 1998, DEC was sold to Compaq.

53. Eric Weiss，《Eloge：AFIPS》，《计算机史年鉴》 13：1（1991）100。

53. Eric Weiss, “Eloge: AFIPS,” Annals of the History of Computing 13: 1 (1991) 100.

第十章

Chapter 10

1.当然，与微软不同，苹果主要是一家硬件公司。但其软件与其硬件创新一样，甚至比硬件创新更能定义其公司身份，而硬件创新通常也相当先进。

1. Unlike Microsoft, of course, Apple is primarily a hardware company. But its software defines its identity as a company as much as, or more than, its hardware innovations, which are often quite advanced as well.

2. MacWrite 具有完整的所见即所得功能，这使它有别于 IBM PC 及其克隆产品，但 MacWrite 用户无法编写超过 10 页的文档。请参阅 Frank Rose 的《伊甸园之西：苹果电脑的纯真时代终结》（纽约：企鹅图书，1989 年），第 11 章。

2. MacWrite had full WYSIWYG capabilities, which set it apart from the IBM PC and its clones, but MacWrite users could not write documents longer than ten pages. See Frank Rose, West of Eden: The End of Innocence at Apple Computer (New York: Penguin Books, 1989), chapter 11.

3.史蒂芬·曼斯 (Stephen Manes) 和保罗·安德鲁斯 (Paul Andrews)，《盖茨：微软大亨如何重塑一个行业并成为美国首富》 (纽约：Doubleday，1993 年)，第 12、13 章。

3. Stephen Manes and Paul Andrews, Gates: How Microsoft’s Mogul Reinvented an Industry, and Made Himself the Richest Man in America (New York: Doubleday, 1993), chapters 12, 13.

4. Susan Lammers 编，《程序员在工作》（华盛顿州雷德蒙德：微软出版社，1986 年）：6-22。

4. Susan Lammers, ed., Programmers at Work (Redmond, WA: Microsoft Press, 1986): 6–22.

5. Lammers，《程序员在工作》，第 207-225 页。微软、Lotus 和 Ashton-Tate 的相对规模数据可在 Martin Campbell-Kelly 的《不仅仅是微软：个人电脑软件行业的成熟，1982-1995 年》中找到，《商业历史评论》（2001 年春季）：第 109 页。

5. Lammers, Programmers at Work, 207–225. Figures for the relative sizes of Microsoft, Lotus, and Ashton-Tate may be found in Martin Campbell-Kelly, “Not Only Microsoft: The Maturing of the Personal Computer Software Industry, 1982–1995,” Business History Review (Spring 2001): 109.

6.所谓的“WIMP”界面：窗口、图标、鼠标和下拉菜单。它胜过集成式单一程序的原因很复杂，可能与社会力量和技术优势同样重要。

6. The so-called “WIMP” interface: Windows, Icons, Mouse, and Pull-down menus. The reasons it prevailed over the integrated single program are complex and probably have as much to do with social forces as with technical superiority.

7.Lammers，《程序员在工作》，13页。

7. Lammers, Programmers at Work, 13.

8. Frederick P. Brooks, Jr.，《人月神话：软件工程论文集》，周年纪念版（马萨诸塞州雷丁：Addison-Wesley，1995 年）：270-271。

8. Frederick P. Brooks, Jr., The Mythical Man-Month: Essays on Software Engineering, anniverary edition (Reading, MA: Addison-Wesley, 1995): 270–271.

9. Michael A. Cusumano 和 Richard W. Selby，《微软的秘密》（纽约：自由出版社，1995 年）：第 36 页，第 269–270 页。作者指出，Excel 3.0 版是第一款采用此策略的 Microsoft 产品。

9. Michael A. Cusumano and Richard W. Selby, Microsoft Secrets (New York: Free Press, 1995): 36, 269–270. The authors notestext1 that version 3.0 of Excel was the first Microsoft product to employ this tactic.

10. G. Pascal Zachary，《引人注目！微软公司争相打造 Windows NT 及下一代操作系统》（纽约：自由出版社，1994 年）：第 1 章。

10. G. Pascal Zachary, Showstopper! The Breakneck Race to Create Windows NT and the Next Generation at Microsoft (New York: Free Press, 1994): chapter 1.

11.养狗的微软程序员可能给狗喂得很好，但这些可怜的动物肯定缺少关爱。在竞争对手网景公司，程序员可以带狗上班，这是该公司引以为豪的一项便利。在软件开发商中， “死亡行军”和“破碎”等术语也是微软及其同行所特有的。

11. Microsoft programmers who owned dogs probably fed them well, but the poor animals must have been starved for affection. At rival Netscape, programmers were allowed to bring their dogs to work, an amenity the company was very proud of. Terms like death march and broken are also peculiar to Microsoft and its peers among software developers.

12. Access 起源于 Fox Software 提供的数据库，微软于 1992 年收购了该数据库。

12. Access had its roots in a database offered by Fox Software, which Microsoft purchased in 1992.

13.有关 PowerPoint 最初开发和随后被微软收购的有趣故事，请参阅 Ian Parker 的《绝对 PowerPoint》，《纽约客》（2001 年 5 月 28 日）：86-87。据 Parker 称，“如今……在美国许多企业中，如果不使用 PowerPoint 参加会议，就会显得不受欢迎，甚至有点装腔作势，就像没穿鞋一样”（第 78 页）。

13. For an amusing story of how PowerPoint was first developed and then acquired by Microsoft, see Ian Parker, “Absolute PowerPoint,” New Yorker (May 28, 2001): 86–87. According to Parker, “Today ... there are great tracts of corporate America where to appear at a meeting without PowerPoint would be unwelcome and vaguely pretentious, like wearing no shoes” (p. 78).

14. Adam Osborne 和 John Dvorak，《Hypergrowth》（加州伯克利：Idthekkethan，1984 年）：162-165；另见 Manes 和 Andrews，《Gates》，第 360-361 页。Martin Campbell-Kelly 指出，Lotus 1-2-3 等程序的定价在 350-550 美元之间，并非基于任何经典经济学模型。很少有经济学理论适用于 PC 软件。

14. Adam Osborne and John Dvorak, Hypergrowth (Berkeley, CA: Idthekkethan, 1984): 162–165; also Manes and Andrews, Gates, 360–361. Martin Campbell-Kelly has pointed out that the pricing of programs like Lotus 1-2-3 in the range of $350–$550 was not based on any classical models of economics. Few economic theories applied to PC software.

15.乔布斯引用了毕加索的名言，据说毕加索曾说过“伟大的艺术家会偷窃”。一些批评乔布斯的人声称毕加索从未说过这句话。似乎这还不够，杰夫·拉斯金声称乔布斯的吹嘘是无稽之谈；毕竟他没有从施乐公司偷东西！只有史蒂夫·乔布斯会因为被指控不是小偷而名誉受损。

15. Jobs quoted Picasso, who allegedly said “Great artists steal.” Some critics of Jobs claim that Picasso never said that. As if that were not enough, Jef Raskin claimed that Jobs’s boast was idle; he did not steal from Xerox after all! Only Steve Jobs can have his reputation tarnished by charges that he is not a thief.

16.随着诉讼的进行，苹果和微软达成协议，授权一种显示和打印字体的技术，该技术后来成为 Windows 的一项基本功能。

16. As that litigation proceeded, Apple and Microsoft entered an agreement to license a technology for displaying and printing type fonts, which later became a fundamental feature of Windows.

17.杰里·卡普兰，《创业：硅谷冒险记》（纽约：霍顿·米夫林，1994 年）：第8、9章。

17. Jerry Kaplan, Start Up: A Silicon Valley Adventure (New York: Houghton Mifflin, 1994): chapters 8, 9.

18.经过几次失败的尝试，包括 GO 的 Penpoint 和苹果的 Newton，手写笔式计算机最终凭借 Palm Pilot 在市场上站稳了脚跟。Palm Pilot 由一家名为 Palm Computing 的初创公司于 1996 年推出。微软推出了自己的手写笔式操作系统，但与 Palm 不兼容，目前，这两款竞争系统的市场份额大致相当。

18. After several false starts including GO’s Penpoint and Apple’s Newton, pen-based computing gained a market foothold with the Palm Pilot, introduced by a start-up called Palm Computing in 1996. Microsoft countered with its own pen-based operating system that was not compatible with the Palm, and at present the two competing systems have about equal market share.

19. Kaplan，《Start Up》，178–181。

19. Kaplan, Start Up, 178–181.

20. Joshua Quittner 和 Michelle Slatalla，《加速网络：Netscape 的内幕故事及其如何挑战微软》（纽约：大西洋月刊出版社，1998 年）：172-174。

20. Joshua Quittner and Michelle Slatalla, Speeding the Net: The Inside Story of Netscape and How It Challenged Microsoft (New York: Atlantic Monthly Press, 1998): 172–174.

21.比尔·盖茨与 Nathan Myhrvold 和 Peter Rinearson 合著《未来之路》（纽约：Viking Press，1995 年）。有关批评的一个例子，请参阅 Mike Wilson 的《上帝与拉里·埃里森的区别：甲骨文公司内部》（纽约：Morrow，1997 年）：第 337-338 页。

21. Bill Gates, with Nathan Myhrvold and Peter Rinearson, The Road Ahead (New York: Viking Press, 1995). For an example of some of the criticism, see Mike Wilson, The Difference between God and Larry Ellison: Inside Oracle Corporation (New York: Morrow, 1997): 337–338.

22.同上。盖茨曾说（第 36 页），DEC 的 Ken Olsen 是“我的英雄，一个遥远的神”。

22. Ibid. At one point (p. 36), Gates says that Ken Olsen of DEC was “a hero of mine, a distant god.”

23.文章发表在几本流行杂志上；例如，参见“微软进军互联网之路”，《商业周刊》（1996 年 7 月 15 日）：封面，56-59。

23. Articles appeared in several popular magazines; see, for example, “Microsoft’s Road to the Internet,” Business Week ( July 15, 1996): cover, 56–59.

24.计算机软件领域的激烈竞争常常导致人们使用战斗和战争的比喻。2001 年 9 月美国遭受恐怖袭击后，这些比喻似乎不再那么无害。

24. The fierce competition in computer software often led to metaphors of combat and war. After the attacks on the United States in September 2001, these metaphors no longer seem as harmless.

25.截至撰写本文时，即 2001 年秋季，作者一直在使用 Netscape 4.7 和 Windows NT 进行办公。我的办公室计算机上没有 IE 图标，我甚至不确定在办公室是否可以使用 IE。

25. As of this writing, in the fall of 2001, the author has been using Netscape 4.7, and Windows NT at work. My office computer does not have the IE icon on it, and I am not even sure if I have access to IE at work at all.

26. Jackon，转引自 Richard B. McKenzie 著《信托受审：微软案如何重塑竞争规则》（剑桥：Perseus Books，2000 年）：第 51 页。

26. Jackon, quoted in Richard B. McKenzie, Trust on Trial: How the Microsoft Case Is Reframing the Rules of Competition (Cambridge: Perseus Books, 2000): 51.

27.本节的研究主要包括观察作者的十几岁的女儿和她的朋友，他们似乎都沉迷于 Hotmail。另请参阅 Po Bronson，《Hot Male》，《Wired》（1998 年 12 月）：166-174。

27. Research for this section consists mainly of observing the author’s preteen daughter and her friends, all of whom seem addicted to Hotmail. See also Po Bronson, “Hot Male,” Wired (December 1998): 166–174.

28.麦肯齐，《信托审判》，53页。

28. McKenzie, Trust on Trial, 53.

29. David Banks，《打破窗户：比尔·盖茨如何搞砸了微软的未来》（纽约：自由出版社，2001 年）。

29. David Banks, Breaking Windows: How Bill Gates Fumbled the Future of Microsoft (New York: Free Press, 2001).

30. Wulf，引自 Christopher Anderson，《数据高速公路的坎坷之路》，《科学》第 260 卷（1993 年 5 月 21 日）：1064–1065 页。

30. Wulf, quoted in Christopher Anderson, “The Rocky Road to a Data Highway,” Science 260 (May 21, 1993): 1064–1065.

31.该网站仍然活跃：请参阅‹ www.shaq.com ›。

31. The site is still active: see ‹www.shaq.com›.

32.关于互联网底层设计理念的最佳表述见于“征求意见稿”（RFC 1958：“互联网架构原则”，网址：www.ietf.org/rfc/rfc1958.txt）。该文写于 1996 年 6 月。1996 年之后，商业流量的激增给设计带来了压力，尽管设计设法应对了，但难度加大了。另见美国国家研究委员会计算机科学与电信委员会的《互联网的成熟》（华盛顿特区：美国国家科学院出版社，2001 年）：第 1 章。

32. The best statement of the Internet’s underlying design philosophy is found in the “Request for Comment” (RFC 1958: “Architectural Principles of the Internet,” found at www.ietf.org/rfc/rfc1958.txt). This was written in June 1996. After 1996 the crush of commercial traffic put strains on the design that it has managed to handle but with greater difficulty. See also National Research Council, Computer Science and Telecommunications Board, The Internet’s Coming of Age (Washington, DC: National Academy Press, 2001): chapter 1.

33.可接受使用政策可以在多个网站上找到，并且也在 Ed Krol 的《整个互联网用户指南和目录》（Sebastopol，CA：O'Reilly & Associates，1992）：353-354 中重印。

33. The Acceptable Use Policy can be found on several Web sites and is also reprinted in Ed Krol, The Whole Internet Users’ Guide and Catalog (Sebastopol, CA: O’Reilly & Associates, 1992): 353–354.

34.因为他以几百票之差落败（佛罗里达州的差距），所以可以说，他犯下的其他错误（不仅仅是这一错误）可能让他输掉了选举。他的采访文本可以在几个网站上找到，其中包括“城市传奇”网站，我依靠该网站进行讨论。

34. Since he lost by a few hundred votes (the margin in Florida), one should say that any number of other missteps he made, not just this one, may have cost him the election. The text of his interview can be found on several Web sites, among them the “Urban Legends” site, which I have relied on for this discussion.

35.重点是我加的。Boucher 的语言被引用于 Stephen Segaller 的《书呆子 2.0.1：互联网简史》（纽约：TV Books，1998 年）：第 296 页。这本书是 PBS 电视网播出的电视连续剧的配套读物。

35. Emphasis mine. Boucher’s language is quoted in Stephen Segaller, Nerds 2.0.1: A Brief History of the Internet (New York: TV Books, 1998): 296. The book was the companion to the television series, broadcast on the PBS network.

36.例如，请参阅戈尔为《科学美国人》 1991 年 9 月特刊撰写的一篇文章，题为“通信、计算机和网络”（1991 年 9 月）：第 150-153 页；以及他于 1994 年 1 月 11 日在加州大学洛杉矶分校“高速公路峰会”上发表的演讲，该演讲可在www.artcontext.com/cal97/superhig.txt上找到，访问时间为 2001 年 12 月 13 日。

36. See, for example, an article Gore wrote for the September 1991 special issue of Scientific American on “Communications, Computers, and Networks” (September 1991): 150–153; also a speech he gave at UCLA on January 11, 1994, to the “Superhighway Summit,” which may be found at www.artcontext.com/cal97/ superhig.txt, accessed December 13, 2001.

37. 20 世纪 80 年代中期，前数字设备公司首席工程师 C. Gordon Bell 被派往国家科学基金会，在那里，他倡导超级计算机中心的概念，并可能是 NREN 计划的主要设计者。

37. In the mid-1980s C. Gordon Bell, formerly Digital Equipment’s chief engineer, was detailed to the NSF, where he championed the notion of supercomputer centers and may have been the key architect of the NREN plan.

38. Anderson，“数据高速公路的坎坷之路”。

38. Anderson, “The Rocky Road to a Data Highway.”

39. Jonathan Coopersmith，《色情、科技与进步》，ICON 4（1998）：94–125。

39. Jonathan Coopersmith, “Pornography, Technology, and Progress,” ICON 4 (1998): 94–125.

40. Peter H. Salus，Casting the Net：从 ARPANET 到 INTERNET 及其他...（马萨诸塞州雷丁：Addison-Wesley，1995 年）：第 222 页。Salus 引用了 1992 年 1 月注册的数字，其中 .edu 域名注册量为 243,020 个，而 .com 域名注册量为 181,361 个，.gov 域名注册量为 46,463 个。

40. Peter H. Salus, Casting the Net: From ARPANET to INTERNET and Beyond ... (Reading, MA: Addison-Wesley, 1995): 222. Salus cites figures for January 1992 of 243,020 registered dot.edu domains, versus 181,361 dot.com and 46,463 dot.gov domains.

41. Joy 曾在私下与 SUN 程序员交流时和公开演讲中使用过这个术语，其中包括 1990 年作者出席的国家航空航天博物馆的一次演讲。

41. Joy used this term both privately to SUN programmers and in public speeches, including one given at the National Air and Space Museum in 1990 attended by the author.

42.历史信息取自亚马逊网站，访问时间为 2000 年 7 月。

42. Historical information was taken from Amazon’s Web site, accessed in July 2000.

43. Robert D. Hof，《人民的公司》，《商业周刊》（电子商务部分）（2001 年 12 月 3 日）：EB-15–21。

43. Robert D. Hof, “People’s Company,” Business Week (e-biz section) (December 3, 2001): EB-15–21.

44. Segaller，Nerds 2.0.1，269。另请参阅 Katie Hafner 的《 The Well：网络社区中关于爱、死亡和真实生活的故事》（纽约：Carroll and Graf，2001 年）。

44. Segaller, Nerds 2.0.1, 269. Also Katie Hafner, The Well: A Story of Love, Death & Real Life in the Seminal Online Community (New York: Carroll and Graf, 2001).

45.本文可在 EFF 的网站上找到：www.eff.org，于 2001 年 11 月访问。Barlow 有自己的主页，可通过 EFF 网站访问，但似乎最近没有更新。

45. The text can be found at the EFF’s Web Site: www.eff.org, accessed in November 2001. Barlow has his own home page, which can be accessed through the EFF’s site, although it appears that it has not been updated recently.

46.作者以这种方式对本章做了一些研究。

46. The author has done some of the research for this chapter in this manner.

47. Robert H. Reid，《网络建筑师：构建商业未来的 1000 天》（纽约：Wiley，1997 年）：第 6 章。

47. Robert H. Reid, Architects of the Web: 1,000 Days That Built the Future of Business (New York: Wiley, 1997): chapter 6.

48.伯纳斯·李，2000 年 4 月 27 日在美国蒙大拿州博兹曼市美国计算机博物馆的演讲；另请参阅他的书《编织万维网》（纽约：HarperCollins，1999 年）。

48. Berners-Lee, speech given at the American Computer Museum, Bozeman, Montana, April 27, 2000; see also his book, Weaving the Web (New York: HarperCollins, 1999).

49.有关 Ted Nelson 的主页，请访问www.sfc.keio.ac.jp/~ted/（2001年 12 月访问）。

49. For Ted Nelson’s home page, see www.sfc.keio.ac.jp/~ted/ (accessed in December 2001).

50.请参阅www.isinet.com/isi/products/citation/sci/（2001年 12 月访问）；Brewster Kahle 的网站于 2001 年在www.archive.orgi 上找到。

50. See www.isinet.com/isi/products/citation/sci/ (accessed in December 2001); Brewster Kahle’s site was found in 2001 at www.archive.orgi.

51.加勒特·哈丁，《公地悲剧》，《科学》 162（1968 年 12 月 13 日）：1243-1248。

51. Garrett Hardin, “The Tragedy of the Commons,” Science 162 (December 13, 1968): 1243–1248.

52.梅特卡夫真的把杂志上的一页都吃掉了，因为他在杂志上预测互联网将因流量增加而崩溃。

52. Metcalfe literally ate the page of the magazine in which he predicted the collapse of the Internet due to increased traffic.

53.例如，请参阅 Nathaniel S. Borenstein 的“曾经的和未来的互联网”，该论文发表于 2001 年 12 月 3-4 日在斯坦福线性加速器中心（SLAC）举行的纪念美国第一个网站成立十周年的研讨会上。概要可参见www.project.slac.stanford.edu/webanniv/nsb.pdfi。

53. See, for example, Nathaniel S. Borenstein, “The Once and Future Internet,” paper presented at a symposium on the occasion of the tenth anniversary of the first U.S. Web site, at the Stanford Linear Accelerator Center (SLAC), December 3–4, 2001. An outline may be found at www.project.slac.stanford.edu/webanniv/ nsb.pdfi.

54.我从多个渠道听到了这个故事，它与围绕西摩·克雷的传说相符，但我无法证实。1988 年 11 月他在佛罗里达州奥兰多举行的会议录像带中没有出现这个内容，但他当时在其他场合也做过基本相同的简报。

54. I have heard this story from several sources, and it is in keeping with the legends surrounding Seymour Cray, but I have been unable to verify it. It does not appear on a videotape of a meeting he gave in Orlando, Florida, in November 1988, but he had given essentially the same briefing at other venues around that time.

55. Glyn Moody，《叛逆的代码：开放源代码革命内幕》（剑桥：Perseus，2001 年）：3、7。

55. Glyn Moody, Rebel Code: Inside the Open Source Revolution (Cambridge: Perseus, 2001): 3, 7.

56. DM Ritchie 和 K. Thompson，《UNIX 分时系统》，《贝尔系统技术期刊》，57/6，第 2 部分（1978 年 7-8 月）：1905-1929，尤其是第 1927 页。

56. D. M. Ritchie and K. Thompson, “The UNIX Time-Sharing System,” Bell System Technical Journal, 57/6, part 2 (July–August 1978): 1905–1929, especially p. 1927.

57.同上，1907-1908 年。

57. Ibid., 1907–1908.

58.最初编写 UNIX 的 PDP-7 的内存容量为 18K，18 位字，或大约 18K 字节。请参阅 Dennis M. Ritchie 的“C 编程语言的发展”，收录于 Thomas J. Bergin 和 Richard G. Gibson 编的《编程语言历史—II》（马萨诸塞州雷丁：Addison-Wesley，1996 年）：第 671-698 页。

58. The PDP-7 on which UNIX was first written had a memory capacity of 18K, 18-bit words, or about 18K bytes. See Dennis M. Ritchie, “The Development of the C Programming Language,” in Thomas J. Bergin and Richard G. Gibson, eds., History of Programming Languages—II (Reading, MA: Addison-Wesley, 1996): 671–698.

59. Manes 和 Andrews, Gates , 147. AT&T 的许可政策对“UNIX”名称进行了限制，因此微软（及其他公司）采用了不同的名称。

59. Manes and Andrews, Gates, 147. AT&T’s licensing policies placed restrictions on the name “UNIX,” hence Microsoft’s (and others’) adopting a different name.

60. Linus Torvalds 和 David Diamond，《只是为了好玩：一个意外革命者的故事》（纽约：HarperCollins，2001 年），第2、3章。

60. Linus Torvalds and David Diamond, Just for Fun: The Story of an Accidental Revolutionary (New York: HarperCollins, 2001), chapters 2, 3.

61. Andrew S. Tanenbaum，《操作系统：设计与实现》（新泽西州恩格尔伍德克利夫斯： Prentice Hall，1987 年）。

61. Andrew S. Tanenbaum, Operating Systems: Design and Implementation (Englewood Cliffs, NJ: Prentice Hall, 1987).

62. Torvalds 和 Diamond，《Just for Fun》，第2、3章。

62. Torvalds and Diamond, Just for Fun, chapters 2, 3.

63.同上，第 61–62 页。

63. Ibid., 61–62.

64. Robert H. Follett 和 Jean E. Sammett，《编程语言标准》，载于 Anthony Ralston、Edwin Reilly 和 David Hemmendinger 编的《计算机科学百科全书》第四版（伦敦：自然出版集团，2000 年）：1466-1470 页。该标准被称为“POSIX”，即“计算机环境的可移植操作系统接口”。

64. Robert H. Follett and Jean E. Sammett, “Programming Language Standards,” in Anthony Ralston, Edwin Reilly, and David Hemmendinger, eds., Encyclopedia of Computer Science, fourth edition (London: Nature Publishing Group, 2000): 1466–1470. The standard was called “POSIX,” for “Portable Operating System Interface for Computer Environments.”

65. Torvalds 和 Diamond，Just For Fun，85。标点和拼写均为原创。部分帖子已保存并存档在网站 ‹Google.com› 上。

65. Torvalds and Diamond, Just For Fun, 85. Punctuation and spelling are original. Some of these postings have been saved and archived on the Web site ‹Google.com›.

66. “口水战”的概念，以及它们是否真正代表了发布此类信息的人的感受，是未来研究的问题，本文不再进一步讨论。脱离上下文来看，称他人的作品“脑残”或说它“糟透了”确实令人震惊，尤其是考虑到欧洲大学对教授的传统尊重。口水战似乎最近已经消亡，尽管它们在网站‹Slashdot.org›上以受限形式存在（见正文）。

66. The notion of “flame wars,” and whether they truly represented the feelings of the persons posting such messages, is a matter for future research and will not be further discussed here. Seen out of context, phrases calling another’s work “brain damaged” or saying that it “sucks” can indeed appear shocking, especially given the traditional respect accorded to professors in European universities. Flame wars seem to have died out recently, although they are alive in a restricted form on the Web site ‹Slashdot.org› (see text).

67. Frederick P. Brooks, Jr.，《人月神话：软件工程论文集》（Reading, MA: Addison Wesley, 1975）。Russ Mitchell 列出了 2001 年 Linux 开发人员“核心圈子”中的 15 人；参见他的文章“Open War”，Wired（2001 年 10 月）：135-139，尤其是第 137 页。他们来自九个国家，几乎都有“日常工作”：他们做其他事情来赚钱。

67. Frederick P. Brooks, Jr., The Mythical Man-Month: Essays on Software Engineering (Reading, MA: Addison Wesley, 1975). Russ Mitchell lists the fifteen people in the “inner circle” of Linux developers as of 2001; see his essay “Open War,” Wired (October 2001): 135–139, especially p. 137. They come from nine countries and nearly all have “day jobs”: they do something else to earn money.

68.布鲁克斯，《人月神话》，80-81。

68. Brooks, The Mythical Man-Month, 80–81.

69. Brooks，《人月神话》周年纪念版（马萨诸塞州雷丁：Addison-Wesley，1995）：203。第 15 章之后添加了新内容。我一直避免依赖这个版本，因为我觉得除了少数例外，它并没有在原版的经典品质上增添多少内容。

69. Brooks, The Mythical Man-Month anniversary edition (Reading, MA: Addison-Wesley, 1995): 203. New material was added after chapter 15. I have avoided relying on this edition, because I feel that, with a few exceptions, it does not add much to the classic qualities of the original.

70.雷蒙德的文章可以在网上找到，但我参考的是已发表的版本，载于《知识、技术与政策》，12/3（1999 年秋季）：23-49。

70. Raymond’s essay is available on the Internet, but I have relied on a published version, in Knowledge, Technology, and Policy, 12/3 (Fall 1999): 23–49.

71.同上，第29页。

71. Ibid., 29.

72. Bryan Pfaffenberger，《恐惧的修辞：信息技术营销中的恐惧、不确定性和怀疑 (FUD)》，《知识、技术和政策》 13/3（2000 年秋季）：78-92。

72. Bryan Pfaffenberger, “The Rhetoric of Dread: Fear, Uncertainty, and Doubt (FUD) in Information Technology Marketing,” Knowledge, Technology, and Policy 13/3 (Fall 2000): 78–92.

73. Torvalds 和 Diamond，《Just for Fun》，87–89。

73. Torvalds and Diamond, Just for Fun, 87–89.

74. Stallman 的个人网页位于 ‹ www.stallman.org/ ›；自由软件基金会的官方网页位于 ‹ www.gnu.ai.mit.edu ›。在线杂志Salon.com一直在刊登 Andrew Leonard 撰写的自由软件运动纪事。作者在 2001-2002 年冬季访问过这些网站，它们可能会发生变化。

74. Stallman’s personal Web page is at ‹www.stallman.org/›; the Free Software Foundation’s official page is at ‹www.gnu.ai.mit.edu›. The on-line magazine Salon.com has been running an ongoing chronicle of the Free Software movement, by Andrew Leonard. These sites were accessed by the author in the winter of 2001–2002, and they may change.

75.穆迪，《反叛守则》，14–19。

75. Moody, Rebel Code, 14–19.

76. 2001 年，搜索引擎 Google.com 恢复并存档了发往 net.unix-wizards 的 Usenet 帖子，本文即取自该帖子。

76. The Usenet posting, to net.unix-wizards, was recovered and archived in 2001 by the search engine Google.com, from which this passage was taken.

77. Richard M. Stallman，《什么是 GNU/Linux 系统？》，《GNU 公告》 1/23（1997）：4-5。

77. Richard M. Stallman, “What Is a GNU/Linux System?”, in GNU’s Bulletin 1/23 (1997): 4–5.

78. Steven Johnson，2002 年 1 月 31 日与作者的私人通信。这些程序的名称通常（但并非总是或始终）用小写字母书写。我尽可能遵循创建它们的人的惯例。

78. Steven Johnson, personal communication to the author, January 31, 2002. The names of these programs are often, but not always or consistently, written in lowercase letters. I have tried to follow the conventions of those who created them wherever possible.

79.同上，第3页。

79. Ibid., 3.

80. Stallman 与 Red Hat 等公司的关系相当融洽，但他反对 O'Reilly & Associates 通过销售作为免费软件手册的书籍来赚钱。根据 Stallman 的说法，这些手册是软件不可或缺的一部分，也应该是免费的。

80. Stallman’s relations with companies like Red Hat are fairly cordial, but he objects to O’Reilly & Associates’ making money by selling books that serve as manuals for free software. According to Stallman, those manuals are an integral part of the software and should be free as well.

81.微软高管詹姆斯·阿尔钦 (James Allchin) 在安德鲁·伦纳德 (Andrew Leonard) 的《生命、自由和对自由软件的追求》中被引述，Salon.com (在线)，2001 年 2 月 15 日。

81. Microsoft executive James Allchin was quoted by Andrew Leonard, “Life, Liberty, and the Pursuit of Free Software,” Salon.com (on-line), February 15, 2001.

82.这个故事的基本梗概取自安德鲁·伦纳德 (Andrew Leonard) 在Salon.com上发表的在线开源历史。

82. The basic outline of this story has been taken from Andrew Leonard’s on-line history of open source, chronicled in Salon.com.

83.同上；John Markoff，“共享软件，IBM 发布邮件计划蓝图”，纽约时报（1998 年 12 月 14 日）：C-5。

83. Ibid.; John Markoff, “Sharing Software, I.B.M. to Release Mail Program Blueprint,” New York Times (December 14, 1998): C-5.

84. Spencer E. Ante，《蓝色巨人押注免费软件》，《商业周刊》（2001 年 12 月 10 日）：第 78-79 页。同一期杂志刊登了一则两页广告，IBM 在其中申明了其对 Linux 的承诺。

84. Spencer E. Ante, “Big Blue’s Big Bet on Free Software,” Business Week (December 10, 2001): 78–79. That same magazine issue carried a two-page ad in which IBM affirmed its commitment to Linux.

85. “IBM 推出首款仅支持 Linux 的大型机”，IBM 新闻稿，2002 年 1 月 25 日。

85. “IBM Unveils First Linux-Only Mainframes,” IBM press release, January 25, 2002.

86. GNOME 代表“Gnu 网络对象模型环境”；KDE 代表“K 桌面环境”。

86. GNOME stands for “Gnu Network Object Model Environment”; KDE for “K Desktop Environment.”

87. Mitchell，《公开战争》。

87. Mitchell, “Open War.”

88.安德森的话被引用于班克斯的《打破窗户》第 26 页，但这句话已经成为民间传说的一部分。

88. Andreesen is quoted in Banks, Breaking Windows, p. 26, but the phrase has become part of common folklore.

89.开源倡导者热切期待 Netscape 浏览器的最新版本，它承诺将采用开源技术。该版本 7.0 最终于 2002 年夏天发布。

89. Open source advocates are eagerly anticipating Netscape’s latest version of its browser, which it promises will be open source. That version, 7.0, was finally released in the summer of 2002.

结论

Conclusion

1. Frederick I. Ordway, III，《2001：太空漫游回顾》，收录于 Eugene M. Emme 主编的《科幻与太空未来，过去与现在》（加利福尼亚州圣地亚哥：美国宇航协会，1982 年）：47-105 页。Ordway 是该片导演 Stanley Kubrick 的顾问。角色/计算机 HAL 的开发是与 IBM、Honeywell、RCA、通用电气以及其他公司和技术专家进行广泛磋商的结果。HAL 的体积似乎比 20 世纪 90 年代的机载计算机大得多，但其对话式用户界面非常接近现代计算机研究人员希望实现的界面。有关我们距离复制 HAL 的距离有多远的评估，请参阅 David G. Stork 主编的《HAL 的遗产：2001 年的计算机是梦想还是现实》（剑桥：麻省理工学院出版社，1997 年）。

1. Frederick I. Ordway, III, “2001: A Space Odyssey in Retrospect,” in Eugene M. Emme, ed., Science Fiction and Space Futures, Past and Present (San Diego, CA: American Astronautical Association, 1982): 47–105. Ordway was a consultant to the film’s director, Stanley Kubrick. The development of the character/ computer HAL was the result of extensive consultations with IBM, Honeywell, RCA, General Electric, and other companies and technical experts. HAL seems to be physically much larger than on-board computers of the 1990s, but in its conversational user interface it is very close to what modern computer researchers hope to attain. For an assessment of how close we are to reproducing HAL, see David G. Stork, ed., HAL’s Legacy: 2001’s Computer as Dream and Reality (Cambridge: MIT Press, 1997).

2. EJ Dijksterhuis，《世界图景的机械化》（牛津：Clarendon Press，1961 年）。

2. E. J. Dijksterhuis, The Mechanization of the World Picture (Oxford: Clarendon Press, 1961).

3.阿兰·图灵，《论可计算数及其在判定问题中的应用》，《伦敦数学学会会刊》第2辑，第42期（1936年）：第230-267页。

3. Alan Turing, “On Computable Numbers, with an Application to the Entscheidungsproblem,” Proceedings London Mathematical Society, Series 2, 42 (1936): 230– 267.

4.我非常感谢奥本大学的W. David Lewis 教授提出这一理念。

4. I am indebted to Professor W. David Lewis of Auburn University for this concept.

5. 《电子学》，1973年10月25日；《时代》，1983年1月3日。

5. Electronics, October 25, 1973; Time, January 3, 1983.

6.例如，这一论点是史密森尼学会 1990 年在美国国家历史博物馆举办的“信息时代”展览的基础。

6. For example, this thesis is the basis for the Smithsonian’s exhibition, “Information Age,” which opened at the National Museum of American History in 1990.

7.例如，请参阅 Clifford Stoll 的《硅蛇油》（纽约：Doubleday，1995 年）。

7. See, for example, Clifford Stoll, Silicon Snake Oil (New York: Doubleday, 1995).

8. Bryan Pfaffenberger，《个人电脑的社会意义，或者为什么个人电脑革命不是革命》，《人类学季刊》第 61 卷（1988 年 1 月）：第 39-47 页。

8. Bryan Pfaffenberger, “The Social Meaning of the Personal Computer, or Why the Personal Computer Revolution was no Revolution,” Anthropological Quarterly 61 ( January 1988): 39–47.

9.梭罗对技术的怀疑态度当然是不同寻常的。最近我听到一位历史学家断言，托马斯·杰斐逊可能是现代计算机尤其是互联网的狂热支持者（David K. Allison，《杰斐逊时代美国的信息革命》，1996 年 5 月 30 日在弗吉尼亚大学为“蒙蒂塞洛回忆录”发表的演讲）。美国国会图书馆将其网站命名为“托马斯”，以纪念杰斐逊。

9. Thoreau’s skepticism about technology was, of course, unusual. Recently I heard a historian assert that Thomas Jefferson would probably have been an enthusiastic proponent of modern computing and especially of the Internet (David K. Allison, “The Information Revolution in Jefferson’s America,” speech given at the University of Virginia for “Monticello Memoirs,” May 30, 1996). The Library of Congress calls its Web site “Thomas” in Jefferson’s honor.

参考书目

Bibliography

Adams Associates。“计算机特性图表”。Datamation （1960 年 11 月/12 月）：14-17 。

Adams Associates. “Computer Characteristics Chart.” Datamation (November/ December 1960): 14–17.

Adams, Charles W. 和 JH Laning Jr.“麻省理工学院自动编码系统：综合、暑期课程和代数。”数字计算机自动编程研讨会(1954): 40–68。

Adams, Charles W., and J. H. Laning Jr. “The MIT Systems of Automatic Coding: Comprehensive, Summer Session, and Algebraic.” Symposium on Automatic Programming of Digital Computers (1954): 40–68.

Aiken, Howard。“自动计算机的未来。” NTF-4 特别版(1956)。

Aiken, Howard. “The Future of Automatic Computing Machinery.” NTF-4 Special Edition (1956).

Allen, John。“TX-0：过去和现在。” 《计算机博物馆报告》，第 8 卷（1984 年春季）：2-11。

Allen, John. “The TX-0: Its Past and Present.” Computer Museum Report, 8 (spring 1984): 2–11.

Allison, David K. “二战以来美国海军的研究与发展。”《军事企业与技术变革》，由 Merritt Roe Smith 编辑，第 289–328 页。剑桥：麻省理工学院出版社，1985 年。

Allison, David K. “U.S. Navy Research and Development Since World War II.” In Military Enterprise and Technological Change, edited by Merritt Roe Smith, 289–328. Cambridge: MIT Press, 1985.

安德森，克里斯托弗。“数据高速公路的坎坷之路。” 《科学》第 260 期（1993 年 5 月 21 日）：1064–1065 页。

Anderson, Christopher. “The Rocky Road to a Data Highway.” Science 260 (May 21, 1993): 1064–1065.

Aspray，William。约翰·冯·诺依曼和现代计算的起源。剑桥：麻省理工学院出版社，1990 年。

Aspray, William. John von Neumann and the Origins of Modern Computing. Cambridge: MIT Press, 1990.

Aspray，William，编辑。计算机之前的计算。艾姆斯，爱荷华州：爱荷华州立大学出版社，1990 年。

Aspray, William, ed. Computing Before Computers. Ames, Iowa: Iowa State University Press, 1990.

Aspray, William 编辑。技术竞争力。纽约：IEEE 出版社，1993 年。

Aspray, William, ed. Technological Competitiveness. New York: IEEE Press, 1993.

Ante, Spencer E. “蓝色巨人押注免费软件。” 《商业周刊》（2001 年 12 月 10 日）：78–79。

Ante, Spencer E. “Big Blue’s Big Bet on Free Software.” Business Week (December 10, 2001): 78–79.

Backus, John。“20 世纪 50 年代的美国编程——一些个人回忆。”《20 世纪计算史》，第 125–135 页。纽约：Academic Press，1980 年。

Backus, John. “Programming in America in the 1950s—Some Personal Recollections.” In A History of Computing in the Twentieth Century, 125–135. New York: Academic Press, 1980.

Baecker, Ronald 和 William AS Buxton。《人机交互读物：一种多学科方法》。洛斯阿尔托斯，加州：Morgan Kaufmann，1987 年。

Baecker, Ronald, and William A. S. Buxton. Readings in Human-Computer Interaction: A Multidisciplinary Approach. Los Altos, CA: Morgan Kaufmann, 1987.

Banks, David。《打破窗户：比尔·盖茨如何搞砸了微软的未来》。纽约：自由出版社，2001 年。

Banks, David. Breaking Windows: How Bill Gates Fumbled the Future of Microsoft. New York: Free Press, 2001.

Barden, William Jr. 《如何购买和使用小型计算机和微型计算机》。印第安纳波利斯，印第安纳州：Howard Sams，1976 年。

Barden, William Jr. How to Buy and Use Minicomputers and Microcomputers. Indianapolis, IN: Howard Sams, 1976.

Basalla, George。“技术的演变。”英国剑桥：剑桥大学出版社，1988 年。

Basalla, George. “The Evolution of Technology.” Cambridge, UK: Cambridge University Press, 1988.

Bashe, Charles J.、Lyle R. Johnson、John H. Palmer 和 Emerson W. Pugh，《IBM 的早期计算机》。剑桥：麻省理工学院出版社，1986 年。

Bashe, Charles J., Lyle R. Johnson, John H. Palmer, and Emerson W. Pugh, IBM’s Early Computers. Cambridge: MIT Press, 1986.

Bauer，FL，《Zuse 和 Rutishauser 之间：中欧数字计算的早期发展》。《二十世纪计算史》，由 J. Howlett、N. Metropolis 和 Gian-Carlo Rota 编辑，第 505–524 页。纽约：Academic Press，1980 年。

Bauer, F. L., “Between Zuse and Rutishauser: The Early Development of Digital Computing in Central Europe.” In A History of Computing in the Twentieth Century, edited by J. Howlett, N. Metropolis, and Gian-Carlo Rota, 505–524. New York: Academic Press, 1980.

Bauer, Roy A.、Emlio Collar 和 Victor Tang。Silverlake项目：IBM 的转型。纽约：牛津大学出版社，1992 年。

Bauer, Roy A., Emlio Collar, and Victor Tang. The Silverlake Project: Transformation at IBM. New York: Oxford University Press, 1992.

Baum, Claude。系统构建者：SDC 的故事。加利福尼亚州圣莫尼卡：系统开发公司，1981 年。

Baum, Claude. The System Builders: The Story of SDC. Santa Monica, CA: System Development Corporation, 1981.

Bell, C. Gordon：“小型和微型产业。” IEEE 计算机17（1984 年 10 月）：14-29。

Bell, C. Gordon: “The Mini and Micro Industries.” IEEE Computer 17 (October 1984): 14–29.

Bell, C. Gordon 和 Allen Newell。计算机结构：阅读和示例。纽约：McGraw Hill，1971 年。

Bell, C. Gordon, and Allen Newell. Computer Structures: Readings and Examples. New York: McGraw Hill, 1971.

Bell, C. Gordon, J. Craig Mudge 和 John McNamara。计算机工程：DEC 对硬件系统设计的看法。贝德福德：马萨诸塞州：Digital Press，1978 年。

Bell, C. Gordon, J. Craig Mudge, and John McNamara. Computer Engineering: A DEC View of Hardware Systems Design. Bedford: MA: Digital Press, 1978.

Bell, C. Gordon 和 John E. McNamara。高科技企业：创业成功指南。马萨诸塞州雷丁：Addison-Wesley，1991 年。

Bell, C. Gordon, and John E. McNamara. High Tech Ventures: The Guide for Entrepreneurial Success. Reading, MA: Addison-Wesley, 1991.

Bell, Trudy E. “20年后获得专利：微处理器的真正之父。” IEEE 研究所14（1990 年 11 月）：1。

Bell, Trudy E. “Patented 20 Years Later: The Microprocessor’s True Father.” The Institute (IEEE) 14 (November 1990): 1.

Beniger, James R. 《控制革命：信息社会的技术和经济起源》。剑桥：哈佛大学出版社，1986 年。

Beniger, James R. The Control Revolution: Technological and Economic Origins of the Information Society. Cambridge: Harvard University Press, 1986.

Bergstein, Harold。“对 Eckert 和 Mauchly 的采访。” Datamation（1962 年 4 月）：25–30。

Bergstein, Harold. “An Interview with Eckert and Mauchly.” Datamation (April 1962): 25–30.

伯克利，埃德蒙。《巨型大脑，或会思考的机器》。纽约：威利出版社，1949 年。

Berkeley, Edmund. Giant Brains, or Machines that Think. New York: Wiley, 1949.

Berners-Lee, Tim。“WWW：过去、现在和未来”，IEEE 计算机29（1996 年 10 月）：69-77。

Berners-Lee, Tim. “WWW: Past, Present, and Future,” IEEE Computer 29 (October 1996): 69–77.

伯纳斯-李，蒂姆。《编织网络》。纽约：HarperCollins，1999 年。

Berners-Lee, Tim. Weaving the Web. New York: HarperCollins, 1999.

伯德，彼得。LEO ：第一台商用计算机。英国沃金厄姆：哈斯勒出版社，1994 年。

Bird, Peter. LEO: the First Business Computer. Wokingham, UK: Hasler Publishing, 1994.

Boehm, Barry。“软件工程。” IEEE Transactions on Computing C25（1976 年 12 月）：1226–1241。

Boehm, Barry. “Software Engineering.” IEEE Transactions on Computing C25 (December 1976): 1226–1241.

Bolter, J. David。《图灵的人：计算机时代的西方文化》。教堂山：北卡罗来纳大学出版社，1984 年。

Bolter, J. David. Turing’s Man: Western Culture in the Computer Age. Chapel Hill: University of North Carolina Press, 1984.

Borenstein, Nathaniel S. “曾经的互联网和未来的互联网。” 2001 年 12 月 3-4 日，在斯坦福线性加速器中心 (SLAC) 举办的纪念美国第一个网站成立十周年的研讨会上发表的论文。大纲可在www.project.slac.stanford.edu/webanniv/nsb.pdfi 上找到。

Borenstein, Nathaniel S. “The Once and Future Internet.” Paper presented at a symposium on the occasion of the tenth anniversary of the first U.S. Web site, at the Stanford Linear Accelerator Center (SLAC), December 3–4, 2001. An outline is available at hwww.project.slac.stanford.edu/webanniv/nsb.pdfi.

Bradley, David。“IBM PC 的诞生。” Byte（1990 年 9 月）：414–420。

Bradley, David. “The Creation of the IBM PC.” Byte (September 1990): 414–420.

布兰德，斯图尔特。“太空战争：计算机迷的狂热生活和象征性死亡。” 《滚石》（1972 年 12 月 7 日），第 50–58 页。

Brand, Stewart. “Spacewar: Fanatic Life and Symbolic Death Among the Computer Bums.” Rolling Stone (7 December 1972), 50–58.

Braun, Ernest 和 Stuart Macdonald。《微型革命：半导体电子学的历史和影响再探——更新和修订的第二版》。第二版。英国剑桥：剑桥大学出版社，1982 年。

Braun, Ernest, and Stuart Macdonald. Revolution in Miniature: The History and Impact of Semiconductor Electronics Re-explored in an Updated and Revised Second Edition. 2nd ed. Cambridge, UK: Cambridge University Press, 1982.

Britcher, Robert N.“卡片、信使和正确性的竞赛。”系统与软件杂志17 (1992): 281–284。

Britcher, Robert N. “Cards, Couriers, and the Race to Correctness.” Journal of Systems and Software 17 (1992): 281–284.

布罗克，杰拉德。《美国计算机行业：市场力量研究》。剑桥：巴林杰出版社，1975 年。

Brock, Gerald. The U.S. Computer Industry: A Study in Market Power. Cambridge: Ballinger, 1975.

Bronson, Po。“热门男性。” 《连线》（1998 年 12 月）：166–174。

Bronson, Po. “Hot Male.” Wired (December 1998): 166–174.

Brooks, Frederick P., Jr. 《人月神话：软件工程论文集》。马萨诸塞州雷丁：Addison-Wesley，1975 年。

Brooks, Frederick P., Jr. The Mythical Man-Month: Essays on Software Engineering. Reading, MA: Addison-Wesley, 1975.

Brooks, Frederick P., Jr. 《人月神话：软件工程论文集》。周年纪念版。马萨诸塞州雷丁：Addison-Wesley，1995 年。

Brooks, Frederick P., Jr. The Mythical Man-Month: Essays on Software Engineering. Anniversary edition. Reading, MA: Addison-Wesley, 1995.

布鲁克斯，约翰。《Go-Go 时代》。纽约：达顿出版社，1984 年。

Brooks, John. The Go-Go Years. New York: Dutton, 1984.

布朗，肯尼斯·A。《发明家在工作；对 16 位著名美国发明家的采访》。华盛顿州雷德蒙德：Tempus Books，1988 年。

Brown, Kenneth A. Inventors at Work; Interviews with 16 Notable American Inventors. Redmond, WA: Tempus Books, 1988.

Brunetti, Cledo 和 Roger W. Curtis。印刷电路技术。通函#468。华盛顿特区：美国国家标准局，1947 年。

Brunetti, Cledo, and Roger W. Curtis. Printed Circuit Techniques. Circular #468. Washington, DC: National Bureau of Standards, 1947.

Buchholz, Werner。“字节一词的起源。” 《计算机历史年鉴》第 3 卷（1981 年）：72。

Buchholz, Werner. “Origin of the Word Byte.” Annals of the History of Computing 3 (1981): 72.

Buchholz，Werner，编辑。规划计算机系统：项目延伸。纽约：麦格劳-希尔，1962 年。

Buchholz, Werner, ed. Planning Computer System: Project Stretch. New York: McGraw-Hill, 1962.

Burke, Steven。“Visicalc 说再见。” 《Infoworld》，1985 年 6 月 24 日，第 20-21 页。

Burke, Steven. “Visicalc Says Goodbye.” Infoworld, 24 June 1985, 20–21.

Burnham, David。《计算机国家的崛起》。纽约：兰登书屋，1983 年。

Burnham, David. The Rise of the Computer State. New York: Random House, 1983.

卡德斯，卡罗琳。《成功的肖像：硅谷先驱者的印象》。加利福尼亚州帕洛阿尔托：蒂奥加，1986 年。

Caddes, Carolyn. Portraits of Success: Impressions of Silicon Valley Pioneers. Palo Alto, CA: Tioga, 1986.

Campbell-Kelly，Martin。“编写 EDSAC：剑桥大学早期的编程活动。”《计算机历史年鉴》第 2 卷（1980 年 1 月）：7-36 页。

Campbell-Kelly, Martin. “Programming the EDSAC: Early Programming Activity at the University of Cambridge.” Annals of the History of Computing 2 ( January 1980): 7–36.

Campbell-Kelly, Martin。ICL ：商业和技术史。牛津：牛津大学出版社，1989 年。

Campbell-Kelly, Martin. ICL: A Business and Technical History. Oxford: Oxford University Press, 1989.

Campbell-Kelly, Martin。“不仅仅是微软：个人电脑软件行业的成熟，1982-1995 年。”《商业历史评论》（2001 年春季）：103-145。

Campbell-Kelly, Martin. “Not Only Microsoft: The Maturing of the Personal Computer Software Industry, 1982–1995.” Business History Review (Spring 2001): 103–145.

保罗·卡罗尔。《大蓝调：IBM 的解体》。纽约：Crown，1994 年。

Carroll, Paul. Big Blues: The Unmaking of IBM. New York: Crown, 1994.

Ceruzzi, Paul。“欧洲计算历史方面。”第 12届欧洲研究会议论文集，内布拉斯加州奥马哈。1987 年 10 月 8-10 日 (1987)：58-66。

Ceruzzi, Paul. “Aspects of the History of European Computing.” Proceedings, 12th European Studies Conference, Omaha, NE. 8–10 October 1987 (1987): 58–66.

Ceruzzi, Paul。超越极限：飞行进入计算机时代。剑桥：麻省理工学院出版社，1989 年。

Ceruzzi, Paul. Beyond the Limits: Flight Enters the Computer Age. Cambridge: MIT Press, 1989.

Chposky, Jan. 《蓝色魔法： IBM 个人电脑背后的人、权力和政治》。纽约：Facts on File，1988 年。

Chposky, Jan. Blue Magic: The People, Power, and Politics Behind the IBM Personal Computer. New York: Facts on File, 1988.

Cohen, Danny。“论圣战与和平恳求。” IEEE 计算机14（1981 年 10 月）：48–54。

Cohen, Danny. “On Holy Wars and a Plea for Peace.” IEEE Computer 14 (October 1981): 48–54.

Cohen, I. Bernard。《科学革命》。剑桥：哈佛大学出版社，贝尔纳普出版社，1985 年。

Cohen, I. Bernard. Revolution in Science. Cambridge: Harvard University Press, Belknap Press, 1985.

计算机及其未来。世界计算机先锋大会。英国兰迪德诺：Richard Williams & Partners，1970 年。

Computers and Their Future. World Computer Pioneer Conference. Llandudno, UK: Richard Williams & Partners, 1970.

Coopersmith, Jonathan。“色情、科技与进步。” ICON 4（1998 年）：94–125。

Coopersmith, Jonathan. “Pornography, Technology, and Progress.” ICON 4 (1998): 94–125.

Cortada, James W.数据处理技术历史词典。康涅狄格州韦斯特波特：格林伍德出版社，1987 年。

Cortada, James W. Historical Dictionary of Data Processing—Technology. Westport, CT: Greenwood Press, 1987.

Cress, Paul、Paul Dirkson 和 J. Wesley Graham。Fortran IV 与 WATFOR 和 WATFIV。新泽西州恩格尔伍德克利夫斯：Prentice Hall，1970 年。

Cress, Paul, Paul Dirkson, and J. Wesley Graham. Fortran IV with WATFOR and WATFIV. Englewood Cliffs, NJ: Prentice Hall, 1970.

Cusumano, Michael A. 和 Richard W. Selby。Microsoft Secrets。纽约：Free Press，1995 年。

Cusumano, Michael A., and Richard W. Selby. Microsoft Secrets. New York: Free Press, 1995.

Daly, RP，“使用 UNIVAC 文件计算机进行集成数据处理”。《西部联合计算机会议论文集》（1956 年）：95-98。

Daly, R. P. “Integrated Data Processing with the UNIVAC File Computer.” Proceedings Western Joint Computer Conference (1956): 95–98.

达比，埃德温。《一切都好起来了：维克托计量仪表公司的成长》。[芝加哥]：维克托计量仪表公司，1968 年。

Darby, Edwin. It All Adds Up: The Growth of Victor Comptometer Corporation. [Chicago]: Victor Comptometer Corporation, 1968.

David, Paul。“理解 QWERTY 的经济学：历史的必要性。”《经济史与现代经济学家》，第 30-49 页。牛津：Blackwell，1986 年。

David, Paul. “Understanding the Economics of QWERTY: The Necessity of History.” In Economic History and the Modern Economist, 30–49. Oxford: Blackwell, 1986.

Davis, EM, WE Harding, RS Schwartz 和 JJ Corning。“固态逻辑技术：多功能高性能微电子技术。” IBM Journal 8（1964 年 4 月）：102–114。

Davis, E. M., W. E. Harding, R. S. Schwartz, and J. J. Corning. “Solid Logic Technology: Versatile, High Performance Microelectronics.” IBM Journal 8 (April 1964): 102–114.

DeLamarter，Thomas。《蓝色巨人：IBM 的权力运用与滥用》。纽约：Dodd，Mead，1986 年。

DeLamarter, Thomas. Big Blue: IBM’s Use and Abuse of Power. New York: Dodd, Mead, 1986.

德兰达，曼努埃尔。《智能机器时代的战争》。剑桥：麻省理工学院出版社，1991 年。

DeLanda, Manuel. War in the Age of Intelligent Machines. Cambridge: MIT Press, 1991.

丹尼斯，迈克尔·A。《国家变迁：1930-1945 年麻省理工学院仪器实验室和约翰霍普金斯大学应用物理实验室技术实践的政治文化》。博士论文，约翰霍普金斯大学，1990 年。

Dennis, Michael A. A Change of State: The Political Cultures of Technological Practice at the MIT Instrumentation Lab and the Johns Hopkins University Applied Physics Laboratory, 1930–1945. Ph.D. diss., Johns Hopkins University, 1990.

迪克森，戴维。《科学的新政治》。芝加哥：芝加哥大学出版社，1984 年。

Dickson, David. The New Politics of Science. Chicago: University of Chicago Press, 1984.

数字设备公司。PDP -11 处理器手册。马萨诸塞州梅纳德：数字设备公司，1981 年。

Digital Equipment Corporation. PDP-11 Processor Handbook. Maynard, MA: Digital Equipment Corporation, 1981.

Dijksterhuis, EJ 《世界图景的机械化》。牛津：克拉伦登出版社，1961 年。

Dijksterhuis, E. J. The Mechanization of the World Picture. Oxford: Clarendon Press, 1961.

Ditlea，Steve。数字熟食店。纽约：Workman，1984 年。

Ditlea, Steve. Digital Deli. New York: Workman, 1984.

Dykstra，DC，“IBM 的信天翁：微型制造商的历史教训。” 《计算机世界》 18（1984 年 12 月 10 日）：134。

Dykstra, D. C. “IBM’s Albatross: A History Lesson for Micro Makers.” Computer-world 18 (10 December 1984): 134.

Eames，Charles 和 Ray，办公室。Glen Fleck 编辑。计算机视角：计算机时代的背景。重印，后记由 Brian Randell 撰写，前言由 I. Bernard Cohen 撰写。剑桥：哈佛大学出版社，1990 年。

Eames, Charles and Ray, office of. Glen Fleck, ed. A Computer Perspective: Background to the Computer Age. Reprint, with an afterword by Brian Randell and foreword by I. Bernard Cohen. Cambridge: Harvard University Press, 1990.

埃克特，华莱士。科学计算中的穿孔卡方法。纽约：哥伦比亚大学，托马斯·沃森天文计算局，1940 年。

Eckert, Wallace. Punched Card Methods in Scientific Computation. New York: Columbia University, The Thomas J. Watson Astronomical Computing Bureau, 1940.

Eckert, Wallace。“IBM 可插拔序列继电器计算器。”数学表和其他计算辅助工具3 (1948)：149–161。

Eckert, Wallace. “The IBM Pluggable Sequence Relay Calculator.” Mathematical Tables and Other Aids to Computation 3 (1948): 149–161.

Ein-Dor, Phillip。“再论格罗施定律。” CACM 28（1985 年 2 月）：142–151。

Ein-Dor, Phillip. “Grosch’s Law Re-revisited.” CACM 28 (February 1985): 142–151.

Eisler, Paul。《我的印刷电路生活》。Mari Williams 编辑并附有注释。宾夕法尼亚州伯利恒：利哈伊大学出版社，1989 年。

Eisler, Paul. My Life with the Printed Circuit. Edited with notes by Mari Williams. Bethlehem, PA: Lehigh University Press, 1989.

Engelbart , Douglas C.“增强人类智力的研究中心”。JCC 33（1968 年秋季）：395–410。

Engelbart, Douglas C. “A Research Center for Augmenting Human Intellect.” JCC 33 (Fall 1968): 395–410.

工程研究协会。高速计算设备。纽约：麦格劳-希尔，1950 年。

Engineering Research Associates. High Speed Computing Devices. New York: McGraw-Hill, 1950.

English, William、Douglas Engelbart 和 Melvyn Berman。“文本操作的显示选择技术。” IEEE 电子人为因素汇刊8（1967 年 3 月）：5-15。

English, William, Douglas Engelbart, and Melvyn Berman. “Display-Selection Techniques for Text Manipulation.” IEEE Transactions on Human Factors in Electronics 8 (March 1967): 5–15.

Evans, Bob O. “SPREAD 报告：IBM System/360 项目的起源。”《计算机历史年鉴》第 5 期（1983 年 1 月）：4-44。Evans, Bob O. “IBM System/360。” 《计算机博物馆报告》（1984 年夏季）：818。

Evans, Bob O. “SPREAD Report: The Origin of the IBM System/360 Project.” Annals of the History of Computing 5 (January 1983): 4–44. Evans, Bob O. “IBM System/360.” Computer Museum Report (Summer 1984): 818.

Ferguson, Eugene S.工程与心灵之眼。剑桥：麻省理工学院出版社，1992 年。Fisher, Franklin, John J. McGowan 和 Joen E. Greenwood。折叠、细长和残缺：经济分析和美国诉 IBM 案。剑桥：麻省理工学院出版社，1983 年。

Ferguson, Eugene S. Engineering and the Mind’s Eye. Cambridge: MIT Press, 1992. Fisher, Franklin, John J. McGowan, and Joen E. Greenwood. Folded, Spindled, and Mutilated: Economic Analysis and U.S. v. IBM. Cambridge: MIT Press, 1983.

Fisher, Franklin、James W. McKie 和 Richard B. Mancke。IBM和美国数据处理行业。纽约：Praeger，1983 年。

Fisher, Franklin, James W. McKie, and Richard B. Mancke. IBM and the U.S. Data Processing Industry. New York: Praeger, 1983.

Fishman, Katherine Davis。《计算机机构》。纽约：Harper & Row，1981 年。

Fishman, Katherine Davis. The Computer Establishment. New York: Harper & Row, 1981.

Flamm, Kenneth。《瞄准计算机：政府支持和国际竞争》。华盛顿特区：布鲁金斯学会，1987 年。

Flamm, Kenneth. Targeting the Computer: Government Support and International Competition. Washington, DC: Brookings Institution, 1987.

Flamm, Kenneth。《创造计算机：政府、工业和高科技》。华盛顿特区：布鲁金斯学会，1988 年。

Flamm, Kenneth. Creating the Computer: Government, Industry, and High Technology. Washington, DC.: Brookings Institution, 1988.

Follett, Robert H 和 Jean E. Sammett。“编程语言标准。”Anthony Ralston、Edwin Reilly 和 David Hemmendinger 编，《计算机科学百科全书》第 4 版。伦敦：自然出版集团，2000 年。

Follett, Robert H, and Jean E. Sammett. “Programming Language Standards.” In Anthony Ralston, Edwin Reilly, and David Hemmendinger, eds., Encyclopedia of Computer Science, 4th ed. London: Nature Publishing Group, 2000.

Frank, Werner。“神话的历史#1。” Datamation（1983 年 5 月）：252–256。

Frank, Werner. “The History of Myth #1.” Datamation (May 1983): 252–256.

Freiberger, Paul。《硅谷之火：个人电脑的诞生》。加州伯克利：McGraw-Hill, Osborne，1984 年。Friedel, Robert。《Zipper：新奇事物的探索》。纽约：WW Norton，1994 年。

Freiberger, Paul. Fire in the Valley: The Making of the Personal Computer. Berkeley, CA: McGraw-Hill, Osborne, 1984. Friedel, Robert. Zipper: An Exploration in Novelty. New York: W.W. Norton, 1994.

Friedlander, Amy。《自然垄断与普遍服务： 1837-1940 年美国通信基础设施中的电话和电报》。弗吉尼亚州雷斯顿国家研究计划公司，1995 年。

Friedlander, Amy. Natural Monopoly and Universal Service: Telephones and Telegraphs in the U.S. Communications Infrastructure, 1837–1940. Reston, VA Corporation for National Research Initiatives, 1995.

Frisch, Ivan T. 和 Howard Frank。“计算机通信：我们如何取得今天的成就。” Proceedings NCC 44 (1975): 109–117。

Frisch, Ivan T., and Howard Frank. “Computer Communications: How We Got Where We Are.” Proceedings NCC 44 (1975): 109–117.

Gardner, W. David。“128 号公路：波士顿的科技温床。” Datamation（1981 年 11 月）：110-118。

Gardner, W. David. “Route 128: Boston’s Hotbed of Technology.” Datamation (November 1981): 110–118.

Gardner, W. David。“Chip Off the Old Bloch。” Datamation（1982 年 6 月）：241–242。

Gardner, W. David. “Chip Off the Old Bloch.” Datamation ( June 1982): 241–242.

Gass, Saul I.“水星计划实时计算与数据流系统。” 《东部 JCC 会议录》，华盛顿特区20（1961 年 12 月 12-14 日）。

Gass, Saul I. “Project Mercury Real-time Computational and Data-flow System.” Proceedings Eastern JCC, Washington, DC 20 (12–14 December 1961).

比尔·盖茨与内森·梅尔沃德和彼得·里尼尔森合著。《未来之路》。纽约：维京出版社，1995 年。

Gates, Bill, with Nathan Myhrvold and Peter Rinearson. The Road Ahead. New York: Viking Press, 1995.

Geraghty, Jim。CICS概念和用途。纽约：McGraw-Hill，1994 年。

Geraghty, Jim. CICS Concepts and Uses. New York: McGraw-Hill, 1994.

Glass, Robert。计算灾难。西雅图：计算趋势，1983 年。

Glass, Robert. Computing Catastrophes. Seattle: Computing Trends, 1983.

Goldberg，Adele，编辑。个人工作站的历史。马萨诸塞州雷丁：Addison-Wesley，1988 年。

Goldberg, Adele, ed. A History of Personal Workstations. Reading, MA: Addison-Wesley, 1988.

Goldstine, Herman H. 《从帕斯卡到冯·诺依曼的计算机》。普林斯顿：普林斯顿大学出版社，1972 年。

Goldstine, Herman H. The Computer from Pascal to von Neumann. Princeton: Princeton University Press, 1972.

Goodman, Seymour E.“苏联计算和技术转让：概述。”世界政治31（1979 年 7 月）：539–570。

Goodman, Seymour E. “Soviet Computing and Technology Transfer: An Overview.” World Politics 31 ( July 1979): 539–570.

Gore, Al.“通信、计算机和网络”。《科学美国人》（1991 年 9 月）：150-153。

Gore, Al. “Communications, Computers, and Networks.” Scientific American (September 1991): 150–153.

Gould, Inc.一家有着深厚根基的年轻公司：Gould , Inc.的历史。伊利诺斯州罗林梅多斯：Gould [1984]。

Gould, Inc. A Young Company with Deep Roots: A History of Gould, Inc. Rolling Meadows, IL: Gould [1984].

Grosch，Herb。计算机：生活中的点滴片段。加利福尼亚州诺瓦托：Third Millennium Books，1991 年。

Grosch, Herb. Computer: Bit Slices from a Life. Novato, CA: Third Millennium Books, 1991.

Guterl, Fred。“设计案例历史：Apple 的 Macintosh。” IEEE Spectrum（1984 年 12 月）：34–43。

Guterl, Fred. “Design Case History: Apple’s Macintosh.” IEEE Spectrum (December 1984): 34–43.

哈夫纳，凯蒂。《井：网络社区中关于爱情、死亡和真实生活的故事》。纽约：卡罗尔和格拉夫出版社，2001 年。

Hafner, Katie. The Well: A Story of Love, Death & Real Life in the Seminal Online Community. New York: Carroll & Graf, 2001.

Hall, Eldon C. “阿波罗导航计算机：设计师的观点。”计算机博物馆报告。波士顿：计算机博物馆，1982 年。

Hall, Eldon C. “The Apollo Guidance Computer: A Designer’s View.” Computer Museum Report. Boston: Computer Museum, 1982.

Hall, Eldon C.月球之旅：阿波罗制导计算机的历史。弗吉尼亚州雷斯顿：AIAA，1996 年。

Hall, Eldon C. Journey to the Moon: the History of the Apollo Guidance Computer. Reston, VA: AIAA, 1996.

Hall，Mark 和 John Barry。Sunburst ：Sun Microsystems 的崛起。芝加哥：当代图书，1990 年。

Hall, Mark, and John Barry. Sunburst: The Ascent of Sun Microsystems. Chicago: Contemporary Books, 1990.

Halstead, WK、JW Leas、JN Marshall 和 EE Minett。“RCA BIZMAC 系统的用途和应用。”西部联合计算机会议论文集(1956)：119–123。

Halstead, W. K., J. W. Leas, J. N. Marshall, and E. E. Minett. “Purpose and Application of the RCA BIZMAC System.” Proceedings Western Joint Computer Conference (1956): 119–123.

汉森，德克。《新炼金术士：硅谷和微电子革命》。波士顿：利特尔，布朗，1982 年。

Hanson, Dirk. The New Alchemists: Silicon Valley and the Microelectronics Revolution. Boston: Little, Brown, 1982.

加勒特·哈丁。“公地悲剧。” 《科学》第 162 期（1968 年 12 月 13 日）：1243–1248 页。

Hardin, Garrett. “The Tragedy of the Commons.” Science 162 (December 13, 1968): 1243–1248.

哈佛大学计算实验室。自动序列控制计算器操作手册。1946 年重印，由 I. Bernard Cohen 作序，Paul E. Ceruzzi 作介绍。剑桥：麻省理工学院出版社，查尔斯·巴贝奇研究所重印系列，第 8 卷，1985 年。

Harvard University Computation Laboratory. A Manual of Operation for the Automatic Sequence Controlled Calculator. 1946, Reprint, with a foreword by I. Bernard Cohen and with an introduction by Paul E. Ceruzzi. Cambridge: MIT Press, Charles Babbage Institute Reprint series, vol. 8, 1985.

哈佛大学计算实验室。大型计算机研讨会论文集。剑桥：哈佛大学出版社，1948 年。

Harvard University Computation Laboratory. Proceedings of a Symposium on Large-Scale Calculating Machinery. Cambridge: Harvard University Press, 1948.

Hendrie, Gardner。“从第一台 16 位迷你计算机到容错计算机。”计算机博物馆报告15（1986 年春季）：7。

Hendrie, Gardner. “From the First 16-bit Mini to Fault-Tolerant Computers.” Computer Museum Report 15 (spring 1986): 7.

Hennessy, John L. 和 David A. Patterson。计算机架构：定量方法。加利福尼亚州圣马特奥：Morgan Kaufmann，1990 年。

Hennessy, John L., and David A. Patterson. Computer Architecture: A Quantitative Approach. San Mateo, CA: Morgan Kaufmann, 1990.

Hennessy, John L. 和 Norman Jouppi。“计算机技术与架构：不断发展的互动。” IEEE Computer 24（1991 年 9 月）：18-29。

Hennessy, John L., and Norman Jouppi. “Computer Technology and Architecture: An Evolving Interaction.” IEEE Computer 24 (September 1991): 18–29.

Heppenheimer，TA，“冯·诺依曼如何指引道路。”美国发明与技术遗产(1990 年秋季)：8-16。

Heppenheimer, T. A. “How von Neumann Showed the Way.” American Heritage of Invention & Technology (fall 1990): 8–16.

Hof, Robert D. “人民的公司。” 《商业周刊》（电子商务部分）（2001 年 12 月 3 日）：EB-15–21。

Hof, Robert D. “The People’s Company.” Business Week (e-biz section) (December 3, 2001): EB-15–21.

Hopper, Grace。“Log Book—ASCC。”马萨诸塞州剑桥，1944 年。史密森学会，美国国家历史博物馆档案馆，Grace Hopper 论文。

Hopper, Grace. “Log Book—ASCC.” Cambridge, MA, 1944. Smithsonian Institution, National Museum of American History Archives, Grace Hopper Papers.

Hopper, Grace。《编译例程。内部备忘录》。费城：雷明顿兰德，1953 年。史密森学会，美国国家历史博物馆档案馆，Grace Hopper 论文。

Hopper, Grace. Compiling Routines. Internal Memorandum. Philadelphia: Remington Rand, 1953. Smithsonian Institution, National Museum of American History Archives, Grace Hopper Papers.

Hopper, Grace。“编译例程。”计算机与自动化2（1953 年 5 月）：1-5。

Hopper, Grace. “Compiling Routines.” Computers and Automation 2 (May 1953): 1–5.

Hord, Michael R. Illiac IV：第一台超级计算机。马里兰州罗克维尔：计算机科学出版社，1982 年。

Hord, Michael R. The Illiac IV: The First Supercomputer. Rockville, MD: Computer Science Press, 1982.

House, Robert W.“OARAC 的可靠性体验。”东部联合计算机会议论文集(1953 年): 43–44。

House, Robert W. “Reliability Experience on the OARAC.” Proceedings Eastern Joint Computer Conference (1953): 43–44.

休斯，托马斯·帕克。《电力网络：西方社会的电气化，1880-1930》。巴尔的摩：约翰霍普金斯大学出版社，1983 年。

Hughes, Thomas Parke. Networks of Power: Electrification in Western Society, 1880– 1930. Baltimore: Johns Hopkins University Press, 1983.

休斯，托马斯·帕克。《美国起源：一个世纪的发明和技术热情，1870-1970》。纽约：企鹅出版社，1989 年。

Hughes, Thomas Parke. American Genesis: A Century of Invention and Technological Enthusiasm, 1870–1970. New York: Penguin, 1989.

Hurd, Cuthbert C.“早期的 IBM 计算机：编辑过的证词。”《计算机历史年鉴》第 3 卷（1981 年 4 月）：163–182。

Hurd, Cuthbert C. “Early IBM Computers: Edited Testimony.” Annals of the History of Computing 3 (April 1981): 163–182.

Johnson，《数据、程序和计算机中的 LR 系统结构》。新泽西州恩格尔伍德克利夫斯：Prentice-Hall，1970 年。

Johnson, L. R. System Structure in Data, Programs, and Computers. Englewood Cliffs, NJ: Prentice-Hall, 1970.

Johnson, Robert。“采访（通用电气，ERMA）。” 《计算机史年鉴》 12：2（1990）：130-137。

Johnson, Robert. “Interview (General Electric, ERMA).” Annals of the History of Computing 12: 2 (1990): 130–137.

卡普兰，杰瑞。创业：硅谷冒险。纽约：霍顿·米夫林，1994 年。

Kaplan, Jerry. Start Up: A Silicon Valley Adventure. New York: Houghton Mifflin, 1994.

Kemeny，John G. 《人与计算机》。纽约：Charles Scribner's Sons，1972 年。

Kemeny, John G. Man and the Computer. New York: Charles Scribner’s Sons, 1972.

肯尼，查尔斯。《驾驭脱缰之马：王安实验室的崛起与衰落》。波士顿：利特尔，布朗，1992 年。

Kenney, Charles. Riding the Runaway Horse: The Rise and Decline of Wang Laboratories. Boston: Little, Brown, 1992.

基德，特蕾西。《新机器的灵魂》。波士顿：利特尔，布朗，1981 年。

Kidder, Tracy. The Soul of a New Machine. Boston: Little, Brown, 1981.

Kidwell, Peggy A. 和 Paul E. Ceruzzi。《数字计算的里程碑：史密森尼图画史》。华盛顿特区：史密森尼出版社，1994 年。

Kidwell, Peggy A., and Paul E. Ceruzzi. Landmarks in Digital Computing: A Smithsonian Pictorial History. Washington, DC: Smithsonian Press, 1994.

Kilburn，T.、DBG Edwards、MJ Lanigan 和 FH Sumner。“单级存储系统”。IRE电子计算机学报。EC-11（1962 年 4 月）：223-235。

Kilburn, T., D. B. G. Edwards, M. J. Lanigan, and F. H. Sumner. “One-Level Storage System.” IRE Transactions on Electronic Computers. EC-11 (April 1962): 223–235.

Kilby, Jack S.“集成电路的发明。” IEEE 电子设备学报23（1976 年 7 月）：648–654。

Kilby, Jack S. “Invention of the Integrated Circuit.” IEEE Transactions on Electronic Devices 23 ( July 1976): 648–654.

Kleiman, Herbert。“集成电路：电子行业产品创新案例研究。”博士论文，乔治华盛顿大学，1966 年。

Kleiman, Herbert. “The Integrated Circuit: A Case Study of Product Innovation in the Electronics Industry.” Ph.D. diss., George Washington University, 1966.

Knapp, Robert H. 和 HB Goodrich。《美国科学家的起源》。1952 年。重印，芝加哥：芝加哥大学出版社，1967 年。

Knapp, Robert H., and H. B. Goodrich. The Origins of American Scientists. 1952. Reprint, Chicago: University of Chicago Press, 1967.

Knuth, Donald。《计算机编程艺术》，3 卷。马萨诸塞州雷丁：Addison-Wesley，1968–1973 年。

Knuth, Donald. The Art of Computer Programming, 3 vols. Reading, MA: Addison-Wesley, 1968–1973.

Knuth, Donald。“冯·诺依曼的第一个计算机程序。” 《计算概览》第 2 卷（1970 年 12 月）：247–260 页。

Knuth, Donald. “Von Neumann’s First Computer Program.” Computing Surveys 2 (December 1970): 247–260.

Knuth, Donald。“编程语言的早期发展。”《二十世纪计算史》第 197-273 页，由N. Metropolis、J. Howlett 和 Gian-Carlo Rota 编辑。纽约：Academic Press，1980 年。

Knuth, Donald. “The Early Development of Programming Languages.” In A History of Computing in the Twentieth Century, 197–273, edited by N. Metropolis, J. Howlett, and Gian-Carlo Rota. New York: Academic Press, 1980.

Krol，编辑。《整个互联网用户指南和目录》。Sebastopol，加州：O'Reilly & Associates，1992 年。

Krol, Ed. The Whole Internet User’s Guide and Catalog. Sebastopol, CA: O’Reilly & Associates, 1992.

Kubie, Elmer。“对第一家软件公司的反思。” 《计算机历史年鉴》 16: 2 (1994): 65–71。

Kubie, Elmer. “Reflections of the First Software Company.” Annals of the History of Computing 16: 2 (1994): 65–71.

Lammers，Susan 编辑。《程序员在工作》。华盛顿州雷德蒙德：微软出版社，1986 年。

Lammers, Susan, ed. Programmers at Work. Redmond, WA: Microsoft Press, 1986.

兰开斯特，唐。TTL食谱。印第安纳波利斯：霍华德·萨姆斯，1974 年。

Lancaster, Don. TTL Cookbook. Indianapolis: Howard Sams, 1974.

Laning、J. Halcomb 和 Neil Zierler。Whirlwind I数学方程式翻译程序。工程备忘录 No. E-364。麻省理工学院。马萨诸塞州剑桥，1954 年。

Laning, J. Halcomb, and Neil Zierler. A Program for Translation of Mathematical Equations for Whirlwind I. Engineering Memorandum No. E-364. MIT. Cambridge, MA, 1954.

Lavington，SH早期英国计算机。马萨诸塞州贝德福德：Digital Press，1980 年。

Lavington, S. H. Early British Computers. Bedford, MA: Digital Press, 1980.

Lavington，SH曼彻斯特计算机史。英国曼彻斯特：NCC 出版社，1975 年。

Lavington, S. H. History of Manchester Computers. Manchester, UK: NCC Publications, 1975.

Levy，Steven。《黑客：计算机革命的英雄》。纽约花园城：Anchor Press/Doubleday，1984 年。

Levy, Steven. Hackers: Heroes of the Computer Revolution. Garden City NY: Anchor Press/Doubleday, 1984.

Licklider，JCR，“人机共生”。IRE Transactions on Human Factors 1（1960 年 3 月）：4-11。

Licklider, J. C. R. “Man-Computer Symbiosis.” IRE Transactions on Human Factors 1 (March 1960): 4–11.

Licklider，JCR，“计算机作为通讯设备。” 《科学与技术》（1968 年 4 月）。

Licklider, J. C. R. “The Computer as a Communications Device.” Science and Technology (April 1968).

Lukoff, Herman。从点到位：电子计算机的个人历史。波特兰，俄勒冈州：机器人出版社，1979 年。

Lukoff, Herman. From Dits to Bits: A Personal History of the Electronic Computer. Portland, OR: Robotics Press, 1979.

Lundstrom, David E.来自 UNIVAC 的几个好人。剑桥：麻省理工学院出版社，1987 年。

Lundstrom, David E. A Few Good Men from UNIVAC. Cambridge: MIT Press, 1987.

McClellan, Stephen T.即将到来的计算机行业大萧条：赢家、输家和幸存者。纽约：Wiley，1984 年。

McClellan, Stephen T. The Coming Computer Industry Shakeout: Winners, Losers, & Survivors. New York: Wiley, 1984.

MacKenzie，CE编码字符集：历史与发展。马萨诸塞州雷丁：Addison-Wesley，1980 年。

MacKenzie, C. E. Coded Character Sets: History & Development. Reading, MA: Addison-Wesley, 1980.

Mackenzie, Donald。《发明准确性：核导弹制导的历史社会学》。剑桥：麻省理工学院出版社，1990 年。

Mackenzie, Donald. Inventing Accuracy: A Historical Sociology of Nuclear Missile Guidance. Cambridge: MIT Press, 1990.

Mackenzie, Donald。“协商算术，构建证明：数学和信息技术社会学。” 《科学社会研究》 23: 1 (1993): 37–65。

Mackenzie, Donald. “Negotiating Arithmetic, Constructing Proof: The Sociology of Mathematics and Information Technology.” Social Studies of Science 23: 1 (1993): 37–65.

麦肯齐，理查德·B。信托案审判：微软案如何重塑竞争规则。剑桥：Perseus，2000 年。

McKenzie, Richard B. Trust on Trial: How the Microsoft Case Is Reframing the Rules of Competition. Cambridge: Perseus, 2000.

McMahon, A. Michal。“计算机与综合体：战后美国技术创新研究”。手稿，NASA 历史办公室，华盛顿特区，1986 年。

McMahon, A. Michal. “The Computer and the Complex: A Study of Technical Innovation in Postwar America.” Manuscript, NASA History Office, Washington, DC, 1986.

“年度机器：计算机的出现。” 《时代》，1983 年 1 月 3 日，封面：14-37。

“Machine of the Year: The Computer Moves In.” Time, 3 January 1983, cover: 14–37.

Maddox, JL、JB O'Toole 和 SY Wong。“TRANSAC S-1000 计算机。”《东部联合计算机会议论文集》（1956 年 12 月）：13-16。

Maddox, J. L., J. B. O’Toole, and S. Y. Wong. “The TRANSAC S-1000 Computer.” Proceedings Eastern Joint Computer Conference (December 1956): 13–16.

Mahoney, Michael。“软件和装配线。”论文发表于电气信息技术技术史研讨会，德意志博物馆，慕尼黑，1990 年。

Mahoney, Michael. “Software and the Assembly Line.” Paper presented at workshop on technohistory of electrical information technology, Deutsches Museum, Munich, 1990.

马内斯，斯蒂芬和保罗·安德鲁斯。盖茨：微软大亨如何重塑行业并成为美国首富。纽约：Doubleday，1993 年。

Manes, Stephen, and Paul Andrews. Gates: How Microsoft’s Mogul Reinvented an Industry, and Made Himself the Richest Man in America. New York: Doubleday, 1993.

Markoff, John。“共享软件，IBM 将发布邮件程序蓝图。”纽约时报（1998 年 12 月 14 日）：C-5。

Markoff, John. “Sharing Software, I.B.M. to Release Mail Program Blueprint.” New York Times (December 14, 1998): C-5.

Metropolis，N.、J. Howlett 和 Gian-Carlo Rota 编。二十世纪计算史。纽约：Academic Press，1980 年。

Metropolis, N., J. Howlett, and Gian-Carlo Rota, eds. A History of Computing in the Twentieth Century. New York: Academic Press, 1980.

Mims, Forrest, III。“Altair 8800 十周年纪念。” 《计算机与电子》（1985 年 1 月）：58–62、81–82。

Mims, Forrest, III. “The Tenth Anniversary of the Altair 8800.” Computers and Electronics (January 1985): 58–62, 81–82.

Misa, Thomas J. “军事需求、商业现实和晶体管的发展，1948-1958 年。”《军事企业和技术变革》，由 Merritt Roe Smith 编辑，第 253-287 页。剑桥：麻省理工学院出版社，1985 年。

Misa, Thomas J. “Military Needs, Commercial Realities, and the Development of the Transistor, 1948–1958.” In Military Enterprise and Technological Change, edited by Merritt Roe Smith, 253–287. Cambridge: MIT Press, 1985.

Mitchell, Russ。“公开战争”。《连线》（2001 年 10 月）：135–139。

Mitchell, Russ. “Open War.” Wired (October 2001): 135–139.

蒙特，汤姆和艾琳·普里蒂金。普里蒂金：治愈美国心灵的人。宾夕法尼亚州艾玛乌斯：罗代尔出版社，1988 年。

Monte, Tom, and Ilene Pritikin. Pritikin: The Man Who Healed America’s Heart. Emmaus, PA: Rodale Press, 1988.

穆迪，格林。叛逆代码：开源革命内幕。剑桥：Perseus，2001 年。

Moody, Glyn. Rebel Code: Inside the Open Source Revolution. Cambridge: Perseus, 2001.

Moore, Gordon E.“微处理器和集成电子技术。” IEEE 64 论文集（1976 年 6 月）：837–841。

Moore, Gordon E. “Microprocessors and Integrated Electronics Technology.” Proceedings of the IEEE 64 ( June 1976): 837–841.

宾夕法尼亚大学摩尔电气工程学院。“电子数字计算机设计理论与技术；摩尔电气工程学院讲座，1946 年 7 月 8 日至 8 月 31 日。”费城：宾夕法尼亚大学，1947-1948 年；重印，剑桥：麻省理工学院出版社，1985 年。

Moore School of Electrical Engineering. University of Pennsylvania. “Theory and Techniques for Design of Electronic Digital Computers; Lectures Given at the Moore School of Electrical Engineering, July 8–August 31, 1946.” Philadelphia: University of Pennsylvania, 1947–1948; Reprint, Cambridge: MIT Press, 1985.

Morris，PR，《世界半导体工业史》。IEE技术史系列12。伦敦：Peter Peregrinus，1990年。

Morris, P. R. A History of the World Semiconductor Industry. IEE History of Technology Series 12. London: Peter Peregrinus, 1990.

Mounce, David C. CICS：一部轻松的编年史。英国赫斯利：IBM，1994 年。

Mounce, David C. CICS: A Light Hearted Chronicle. Hursley, UK: IBM, 1994.

默里，查尔斯。《超人：西摩·克雷和超级计算机背后的技术奇才的故事》。纽约：约翰·威利，1997 年。

Murray, Charles. The Supermen: The Story of Seymour Cray and the Technical Wizards behind the Supercomputer. New York: John Wiley, 1997.

Murray, Francis J.数学机器。第 1 卷，数字计算机。纽约：哥伦比亚大学出版社，1961 年。

Murray, Francis J. Mathematical Machines. Vol. 1, Digital Computers. New York: Columbia University Press, 1961.

Myer, TH 和 IE Sutherland。“论显示处理器的设计。” 《ACM 通讯》第 11 卷（1968 年 6 月）：410–414。

Myer, T. H., and I. E. Sutherland. “On the Design of Display Processors.” Communications of the ACM 11 ( June 1968): 410–414.

美国国家研究委员会、计算机科学和电信委员会。《互联网的成熟》。华盛顿特区：美国国家科学院出版社，2001 年。

National Research Council, Computer Science and Telecommunications Board. The Internet’s Coming of Age. Washington, DC: National Academy Press, 2001.

Naur, Peter 和 Brian Randell。软件工程；北约科学委员会主办的会议报告，1968 年10月 7-11 日。德国加米施：北约，1969 年。

Naur, Peter, and Brian Randell. Software Engineering; Report on a Conference Sponsored by the NATO Science Committee, 7–11 October, 1968. Garmisch, Germany: NATO, 1969.

纳尔逊，西奥多。计算机图书馆。印第安纳州南本德：泰德·纳尔逊，1974 年。

Nelson, Theodor. Computer Lib. South Bend, IN: Ted Nelson, 1974.

诺布尔，大卫 F. 《生产力》。纽约：克诺夫出版社，1984 年。

Noble, David F. Forces of Production. New York: Knopf, 1984.

Norberg, Arthur。“二十世纪初的高科技计算：商业和政府中的打孔卡机器。” 《科技与文化》第 31 卷（1990 年 10 月）：753–779 页。

Norberg, Arthur. “High Technology Calculation in the Early Twentieth Century: Punched Card Machinery in Business and Government.” Technology and Culture 31 (October 1990): 753–779.

Norberg，Arthur，Judy O'Neill 和 Kerry Freedman。国防高级研究计划局信息处理技术办公室的历史。CBI 私人印刷的打字稿。明尼阿波利斯，明尼苏达州：查尔斯·巴贝奇研究所，1992 年。

Norberg, Arthur, Judy O’Neill, and Kerry Freedman. A History of the Information Processing Techniques Office of the Defense Advanced Research Projects Agency. Privately printed typescript from CBI. Minneapolis, MN: Charles Babbage Institute, 1992.

Norman, Donald A.“UNIX 的麻烦”。Datamation （1981 年 11 月）：139–150 。

Norman, Donald A. “The Trouble with UNIX.” Datamation (November 1981): 139–150.

Noyce, Robert 和 Marcian Hoff。“英特尔微处理器设计史。” IEEE Micro 1（1981 年 2 月）：8-22。

Noyce, Robert, and Marcian Hoff. “A History of Microprocessor Design at Intel.” IEEE Micro 1 (February 1981): 8–22.

Noyes, T. 和 WE Dickenson。“磁盘随机存取存储器的工程设计。”西部联合计算机会议论文集（ 1956 年 2 月 7 日）：42-44。

Noyes, T., and W. E. Dickenson. “Engineering Design of a Magnetic Disk Random Access Memory.” Proceedings, Western Joint Computer Conference (7 February 1956): 42–44.

Nyce, James 和 Paul Kahn 编。从 Memex 到超文本：Vannevar Bush 和心灵机器。波士顿：Academic Press，1991 年。

Nyce, James, and Paul Kahn, eds. From Memex to Hypertext: Vannevar Bush and the Mind’s Machine. Boston: Academic Press, 1991.

O'Neill, Judy E.《通过分时和网络实现交互式计算的演进》。博士论文，明尼苏达大学，1992 年。

O’Neill, Judy E. The Evolution of Interactive Computing Through Time-Sharing and Networking. Ph.D. diss., University of Minnesota, 1992.

Orchard-Hays, William。“编程系统的演进。” IRE Proceedings 49（1961 年 1 月）：283–295。

Orchard-Hays, William. “The Evolution of Programming Systems.” IRE Proceedings 49 (January 1961): 283–295.

Osborn, Roddy F.“GE 和 UNIVAC：驾驭高速计算机。” 《哈佛商业评论》（1954 年 7-8 月）：99-107。

Osborn, Roddy F. “GE and UNIVAC: Harnessing the High-Speed Computer.” Harvard Business Review ( July–August 1954): 99–107.

奥斯本，亚当和约翰·德沃夏克。超速增长。加州伯克利：Idthekkethan，1984 年。

Osborne, Adam, and John Dvorak. Hypergrowth. Berkeley, CA: Idthekkethan, 1984.

Padegs, A.“System/360 及其他。” IBM 研究与开发杂志25（1981 年 9 月）：377–390。

Padegs, A. “System/360 and Beyond.” IBM Journal of Research and Development 25 (September 1981): 377–390.

Pake, George。“施乐帕洛阿尔托研究中心的研究：创始人的评估。” IEEE Spectrum（1985 年 10 月）：54-75。

Pake, George. “Research at XEROX PARC: A Founder’s Assessment.” IEEE Spectrum (October 1985): 54–75.

Parker, Ian。“Absolute PowerPoint。”纽约客（2001 年 5 月 28 日）：86–87。

Parker, Ian. “Absolute PowerPoint.” New Yorker (May 28, 2001): 86–87.

Patterson, David A.“精简指令集计算机。” CACM 28: 1 (1985): 8–21。

Patterson, David A. “Reduced Instruction Set Computers.” CACM 28: 1 (1985): 8–21.

Pearson, Jamie。《数字化工作》。马萨诸塞州贝德福德：Digital Press，1992 年。

Pearson, Jamie. Digital at Work. Bedford, MA: Digital Press, 1992.

Petroski，Henry。《铅笔：设计与环境的历史》。纽约：Knopf，1990 年。

Petroski, Henry. The Pencil: A History of Design and Circumstance. New York: Knopf, 1990.

Pfaffenberger, Bryan。“个人电脑的社会意义，或者为什么个人电脑革命不是革命。” 《人类学季刊》第 61 期（1988 年 1 月）：39–47。

Pfaffenberger, Bryan. “The Social Meaning of the Personal Computer, or Why the Personal Computer Revolution Was No Revolution.” Anthropological Quarterly 61 ( January 1988): 39–47.

Pfaffenberger, Bryan。“恐惧的修辞：信息技术营销中的恐惧、不确定性和怀疑 (FUD)。”知识、技术和政策13/3（2000 年秋季）：78–92。

Pfaffenberger, Bryan. “The Rhetoric of Dread: Fear, Uncertainty, and Doubt (FUD) in Information Technology Marketing.” Knowledge, Technology, and Policy 13/3 (fall 2000): 78–92.

Phister, Montgomery Jr.“Quotron II：一种用于传输股票市场数据的早期多程序多处理器。”计算机史年鉴11: 2 (1989): 109–126。

Phister, Montgomery Jr. “Quotron II: An Early Multiprogrammed Multiprocessor for the Communication of Stock Market Data.” Annals of the History of Computing 11: 2 (1989): 109–126.

Pollack, S.计算机科学研究。华盛顿特区：美国数学协会，1982 年。

Pollack, S. Studies in Computer Science. Washington, DC: Mathematical Association of America, 1982.

Prasad, NS IBM 大型机：架构与设计。纽约：McGraw Hill，1989 年。

Prasad, N. S. IBM Mainframes: Architecture and Design. New York: McGraw Hill, 1989.

Pugh, Emerson。《塑造一个行业的记忆》。剑桥：麻省理工学院出版社，1984 年。

Pugh, Emerson. Memories that Shaped an Industry. Cambridge: MIT Press, 1984.

Pugh, Emerson。《创建 IBM：塑造行业及其技术》。剑桥：麻省理工学院出版社，1995 年。

Pugh, Emerson. Building IBM: Shaping an Industry and Its Technology. Cambridge: MIT Press, 1995.

Pugh, Emerson W.、Lyle R. Johnson 和 John H. Palmer。IBM的 360 和早期 370 系统。剑桥：麻省理工学院出版社，1991 年。

Pugh, Emerson W., Lyle R. Johnson, and John H. Palmer. IBM’s 360 and Early 370 Systems. Cambridge: MIT Press, 1991.

Quittner，Joshua 和 Michelle Slatalla。《加速网络：Netscape 的内幕故事及其如何改变微软》。纽约：Atlantic Monthly Press，1998 年。

Quittner, Joshua, and Michelle Slatalla. Speeding the Net: The Inside Story of Netscape and How It Changed Microsoft. New York: Atlantic Monthly Press, 1998.

Randell，Brian，编辑。数字计算机的起源：精选论文。第二版。计算机科学文本和专著。柏林：Springer-Verlag，1975 年。

Randell, Brian, ed. The Origins of Digital Computers: Selected Papers. 2nd ed. Texts and Monographs in Computer Science. Berlin: Springer-Verlag, 1975.

Raymond, Eric。“大教堂和集市。” 《知识、技术与政策》 12/3（1999 年秋季）：23–49。

Raymond, Eric. “The Cathedral and the Bazaar.” Knowledge, Technology, and Policy 12/3 (fall 1999): 23–49.

Rebello, Kathy。“微软进军互联网之路。” 《商业周刊》（1996 年 7 月 15 日）：封面，56-59。

Rebello, Kathy. “Microsoft’s Road to the Internet.” Business Week ( July 15, 1996): cover, 56–59.

Redmond，Kent C. 和 Thomas M. Smith。旋风计划：先锋计算机的历史。贝德福德：数字出版社，1980 年。

Redmond, Kent C. and Thomas M. Smith. Project Whirlwind: The History of a Pioneer Computer. Bedford: Digital Press, 1980.

Reid, Robert H.网络建筑师：构建商业未来的 1000 天。纽约：Wiley，1997 年。

Reid, Robert H. Architects of the Web: 1,000 Days That Built the Future of Business. New York: Wiley, 1997.

Reid, TR 《芯片：两个美国人如何发明微芯片并引发革命》。纽约：西蒙与舒斯特出版社，1985 年。

Reid, T. R. The Chip: How Two Americans Invented the Microchip and Launched a Revolution. New York: Simon & Schuster, 1985.

里夫金，格伦和乔治·哈拉。《终极企业家：肯·奥尔森和数字设备公司的故事》。芝加哥：当代图书出版社，1988 年。

Rifkin, Glenn, and George Harrar. The Ultimate Entrepreneur: The Story of Ken Olsen and Digital Equipment Corporation. Chicago: Contemporary Books, 1988.

Ritchie, Dennis M. “UNIX 分时系统：回顾”。《贝尔系统技术期刊》 57（1978 年 7-8 月）：1947-1969 年。

Ritchie, Dennis M. “UNIX Time-Sharing System: A Retrospective.” The Bell System Technical Journal 57 ( July–August 1978): 1947–1969.

Ritchie, Dennis M. 和 K. Thompson。“UNIX 分时系统”。《贝尔系统技术期刊》 57（1978 年 7-8 月）：1905-1929 年。

Ritchie, Dennis M., and K. Thompson. “UNIX Time-Sharing System.” The Bell System Technical Journal 57 ( July–August 1978): 1905–1929.

Ritchie, Dennis M. “C 编程语言的发展。”摘自 Thomas J. Bergin 和 Richard G. Gibson 编著的《编程语言历史—II》。马萨诸塞州雷丁：Addison-Wesley，1996 年。

Ritchie, Dennis M. “The Development of the C Programming Language.” In Thomas J. Bergin and Richard G. Gibson, eds., History of Programming Languages—II. Reading, MA: Addison-Wesley, 1996.

Roberts, H. Edward 和 William Yates。“独家！Altair 8800：有史以来最强大的小型计算机项目 — 制造成本不到 400 美元。” 《大众电子》（1975 年 1 月）：封面，第 33-38 页。

Roberts, H. Edward, and William Yates. “Exclusive! Altair 8800: the Most Powerful Minicomputer Project Ever Presented—Can be Built for Under $400.” Popular Electronics (January 1975): cover, 33–38.

Robinson，LP，“30-201 型电子数字计算机”。中等价位商用通用数字计算机研讨会。美国海军，海军研究办公室，华盛顿特区（1952 年 5 月 14 日）：第 31-36 页。

Robinson, L. P. “Model 30-201 Electronic Digital Computer.” Symposium on Commercially-available General-purpose Digital Computers of Moderate Price. U.S. Navy, Office of Naval Research, Washington, DC (14 May 1952): 31–36.

Rodgers, William。《思考：沃森一家和 IBM 的传记》。纽约：Stein & Day，1969 年。

Rodgers, William. Think: A Biography of the Watsons and IBM. New York: Stein & Day, 1969.

罗斯，弗兰克。《伊甸园之西：苹果电脑的纯真时代终结》。纽约：企鹅图书，1989 年。

Rose, Frank. West of Eden: The End of Innocence at Apple Computer. New York: Penguin Books, 1989.

Rosen, Saul。“电子计算机：历史调查”。《计算调查》 1（1969 年 3 月）：7-36。

Rosen, Saul. “Electronic Computers: A Historical Survey.” Computing Surveys 1 (March 1969): 7–36.

Rosen, Saul 编辑。编程系统和语言。纽约：McGraw-Hill，1967 年。

Rosen, Saul, ed. Programming Systems and Languages. New York: McGraw-Hill, 1967.

鲁蒂豪瑟，海因茨。 “在编程器中自动执行清洁计划。” ZAMP 3 (1952): 312–313。鲁蒂豪瑟，海因茨。 “在编程器中执行编程计划。”来自 ETH 3 应用数学研究所的论文(1952)。

Rutishauser, Heinz. “Automatische Rechenplanfertigung bei Programmgesteuerten Rechenmaschinen.” ZAMP 3 (1952): 312–313. Rutishauser, Heinz. “Rechenplanfertigung bei programmgesteuerten Rechenmaschinen.” Mitteilungen aus dem Institut fu¨r Angewandte Mathematik der ETH 3 (1952).

Salus, Peter。《UNIX 的四分之一世纪》。马萨诸塞州雷丁：Addison-Wesley，1994 年。

Salus, Peter. A Quarter Century of UNIX. Reading, MA: Addison-Wesley, 1994.

Salus, Peter。《撒网：从 ARPANET 到 INTERNET 及以后》 ... 马萨诸塞州雷丁：Addison-Wesley，1995 年。

Salus, Peter. Casting the Net: From ARPANET to INTERNET and Beyond... Reading, MA: Addison-Wesley, 1995.

Sammet, Jean。编程语言：历史和基础知识。新泽西州恩格尔伍德克利夫斯：普伦蒂斯霍尔，1969 年。

Sammet, Jean. Programming Languages: Htstory and Fundamentals. Englewood Cliffs, NJ: Prentice-Hall, 1969.

施耐德，本·罗斯。《计算机世界之旅》。马萨诸塞州雷丁：Addison-Wesley，1974 年。

Schneider, Ben Ross. Travels in Computerland. Reading, MA: Addison-Wesley, 1974.

斯科特，玛丽莲和罗伯特·霍夫曼。“水星编程系统。” 《东部 JCC 会议录》第 20 卷（1961 年 12 月 12-14 日）：47-53。

Scott, Marilyn, and Robert Hoffman. “The Mercury Programming System.” Proceedings Eastern JCC 20 (12–14 December 1961): 47–53.

Segaller，Stephen。《书呆子 2.0.1：互联网简史》。纽约：TV Books，1998 年。

Segaller, Stephen. Nerds 2.0.1: A Brief History of the Internet. New York: TV Books, 1998.

Siewiorek，Daniel P.，C. Gordon Bell 和 Allen Newell。计算机结构：原理和示例。纽约：McGraw-Hill，1982 年。

Siewiorek, Daniel P., C. Gordon Bell, and Allen Newell. Computer Structures: Principles and Examples. New York: McGraw-Hill, 1982.

西蒙，赫伯特。《人工智能科学》。第二版。剑桥：麻省理工学院出版社，1981 年。

Simon, Herbert. The Sciences of the Artificial. 2nd ed. Cambridge: MIT Press, 1981.

斯莱特，罗伯特。《硅中的肖像》。剑桥：麻省理工学院出版社，1992 年。

Slater, Robert. Portraits in Silicon. Cambridge: MIT Press, 1992.

Smith, David C.、Charles Irby、Ralph Kimball 和 Bill Verplank。“设计星型用户界面。” Byte（1982 年 4 月）：242–282。

Smith, David C., Charles Irby, Ralph Kimball, and Bill Verplank. “Designing the Star User Interface.” Byte (April 1982): 242–282.

史密斯，道格拉斯和罗伯特·亚历山大。《摸索未来：施乐公司如何发明并忽视了第一台个人电脑》。纽约：威廉莫罗出版社，1988 年。

Smith, Douglas, and Robert Alexander. Fumbling the Future: How Xerox Invented, Then Ignored, the First Personal Computer. New York: William Morrow, 1988.

Smith, Richard E.，“计算机架构的历史概述”。《计算机史年鉴》 10: 4 (1989): 277–303。

Smith, Richard E. “A Historical Overview of Computer Architecture.” Annals of the History of Computing 10: 4 (1989): 277–303.

Snively, George。“通用电气进军计算机业务。” 《计算机历史年鉴》 10: 1 (1988): 74–78。

Snively, George. “General Electric Enters the Computer Business.” Annals of the History of Computing 10: 1 (1988): 74–78.

Snyder, Samuel S.“美国密码组织对数字计算机行业的影响。”系统与软件杂志1 (1979): 87–102。

Snyder, Samuel S. “Influence of U.S. Cryptologic Organizations on the Digital Computer Industry.” Journal of Systems and Software 1 (1979): 87–102.

索贝尔，罗伯特。IBM ：转型中的巨人。纽约：Bantam，1981 年。

Sobel, Robert. IBM: Colossus in Transition. New York: Bantam, 1981.

Stallman, Richard M.“什么是 GNU/Linux 系统？” GNU 公告1/23（1997）：4–5。

Stallman, Richard M. “What Is a GNU/Linux System?” GNU’s Bulletin 1/23 (1997): 4–5.

斯特恩，南希。从 ENIAC 到 UNIVAC：对 Eckert-Mauchly 计算机的评价。马萨诸塞州贝德福德：数字出版社，1981 年。

Stern, Nancy. From ENIAC to UNIVAC: An Appraisal of the Eckert-Mauchly Computers. Bedford, MA: Digital Press, 1981.

斯托尔，克利福德。硅蛇油。纽约：Doubleday，1995 年。

Stoll, Clifford. Silicon Snake Oil. New York: Doubleday, 1995.

Stockdale, Ian。“供应商、国际标准社区采用 Fortran 90。” NAS News（NASA 艾姆斯研究中心） 7（1992 年 9 月）：1-8。

Stockdale, Ian. “Vendors, International Standards Community Adopt Fortran 90.” NAS News (NASA Ames Research Center) 7 (September 1992): 1–8.

Stork, David G.，编辑。HAL的遗产：2001 年的计算机是梦想还是现实。剑桥：麻省理工学院出版社，1997 年。

Stork, David G., ed. HAL’s Legacy: 2001’s Computer as Dream and Reality. Cambridge: MIT Press, 1997.

Tanenbaum, Andrew S.操作系统：设计与实现。新泽西州恩格尔伍德克利夫斯：Prentice Hall，1987 年。

Tanenbaum, Andrew S. Operating Systems: Design and Implementation. Englewood Cliffs, NJ: Prentice Hall, 1987.

托夫勒，阿尔文。《第三次浪潮》。纽约：莫罗出版社，1980 年。

Toffler, Alvin. The Third Wave. New York: Morrow, 1980.

Tomayko, James。太空飞行中的计算机：NASA 经验。NASA 承包商报告 182505，华盛顿特区：NASA，1988 年。

Tomayko, James. Computers in Spaceflight: The NASA Experience. NASA Contractor Report 182505, Washington, DC: NASA, 1988.

Tomayko, James。“北约软件工程会议。” 《计算史年鉴》 11: 2 (1989): 132–143。

Tomayko, James. “The NATO Software Engineering Conferences.” Annals of the History of Computing 11: 2 (1989): 132–143.

Torvalds，Linus 和 David Diamond。《只是为了好玩：一个偶然革命者的故事》。纽约：HarperCollins，2001 年。

Torvalds, Linus, and David Diamond. Just for Fun: The Story of an Accidental Revolutionary. New York: HarperCollins, 2001.

图灵，艾伦。“论可计算数及其在Entschei dungsproblem中的应用。”伦敦数学学会会刊2：42（1936）：230–267。

Turing, Alan. “On Computable Numbers, with an Application to the Entschei dungsproblem.” Proceedings, London Mathematical Society 2: 42 (1936): 230–267.

Turkle, Sherry。《第二自我：计算机与人类精神》。纽约：西蒙与舒斯特出版社，1984 年。

Turkle, Sherry. The Second Self: Computers and the Human Spirit. New York: Simon & Schuster, 1984.

“TX-0：它的过去和现在。”计算机博物馆报告8（1984 年春季）：2-11。

“The TX-0: Its Past and Present.” Computer Museum Report 8 (spring 1984): 2–11.

美国空军。AESOP ：一种通用的实时直接访问管理信息系统方法。MTP-33；AD634371。马萨诸塞州贝德福德：MITRE 公司，1966 年。

U.S. Air Force. AESOP: A General Purpose Aproach to Real-Time Direct Access Management Information Systems. MTP-33; AD634371. Bedford, MA: MITRE Corporation, 1966.

美国海军数学顾问小组。数字计算机自动编程研讨会。华盛顿特区：美国海军，1954 年。

U.S. Naval Mathematics Advisory Panel. Symposium on Automatic Programming for Digital Computers. Washington, DC: US Navy, 1954.

美国海军研究办公室。数字计算机设备管理方面研讨会。华盛顿特区，1953 年 3 月 30 日。

U.S. Office of Naval Research. Symposium on Managerial Aspects of Digital Computer Installations. Washington, DC, March 30, 1953.

Van Deusen, Edmund。“电子技术走向现代化。”财富（1955 年 6 月）132–136、148。

Van Deusen, Edmund. “Electronics Goes Modern.” Fortune ( June 1955) 132– 136, 148.

Veit, Stan。Stan Veit 的个人电脑史。北卡罗来纳州阿什维尔：WorldComm，1993 年。

Veit, Stan. Stan Veit’s History of the Personal Computer. Asheville, NC: WorldComm, 1993.

王安。《教训》。马萨诸塞州雷丁：Addison-Wesley，1986 年。

Wang, An. Lessons. Reading, MA: Addison-Wesley, 1986.

Warren, Jim。“个人计算：面向计算机专业人员的概述。” NCC Proceedings 46 (1977): 493–498。

Warren, Jim. “Personal Computing: An Overview for Computer Professionals.” NCC Proceedings 46 (1977): 493–498.

沃森，小托马斯。《父亲、儿子和公司：我在 IBM 的生活及以后》。纽约：Bantam Books，1990 年。

Watson, Thomas Jr. Father, Son & Co.: My Life at IBM and Beyond. New York: Bantam Books, 1990.

Weiss，Eric，编辑。《计算机使用：基础知识》。纽约：计算机使用公司，1969 年。

Weiss, Eric, ed. Computer Usage: Fundamentals. New York: Computer Usage Corporation, 1969.

Weizenbaum, Joseph。计算机能力与人类理性。旧金山：WH Freeman，1976 年。

Weizenbaum, Joseph. Computer Power and Human Reason. San Francisco: W. H. Freeman, 1976.

Wexelblat，Richard，编辑。《编程语言的历史》。纽约：Academic Press，1981 年。

Wexelblat, Richard, ed. History of Programming Languages. New York: Academic Press, 1981.

Wilkes, Maurice。“设计自动计算机的最佳方法”（1951 年曼彻斯特首次会议上发表的论文）。《计算机设计开发：主要论文》，Earl Swartzlander 编辑，新泽西州罗谢尔公园：Hayden Book Company，1976 年：266-270。

Wilkes, Maurice. “The Best Way to Design an Automatic Calculating Machine” (paper presented at the Manchester Inaugural Conference, 1951). In Computer Design Development: Principal Papers, edited by Earl Swartzlander, Rochelle Park, NJ: Hayden Book Company, 1976: 266–270.

Wilkes, Maurice V.自动数字计算机。伦敦：Methuen，1956 年。

Wilkes, Maurice V. Automatic Digital Computers. London: Methuen, 1956.

Wilkes, Maurice V, David J. Wheeler 和 Stanley Gill。《电子数字计算机程序的编写》。剑桥：Addison-Wesley，1951 年。

Wilkes, Maurice V, David J. Wheeler, and Stanley Gill. The Preparation of Programs for an Electronic Digital Computer. Cambridge: Addison-Wesley, 1951.

Wilson, Mike。上帝与拉里·埃里森的区别：甲骨文公司内部。纽约：Morrow，1997 年。Wolff, Michael F. “集成电路的起源。” IEEE Spectrum（1976 年 8 月）：45-53。

Wilson, Mike. The Difference between God and Larry Ellison: Inside Oracle Corporation. New York: Morrow, 1997. Wolff, Michael F. “The Genesis of the IC.” IEEE Spectrum (August 1976): 45–53.

Wolfe, Tom。“罗伯特·诺伊斯的修补”。《时尚先生》（1983 年 12 月）：346–374。

Wolfe, Tom. “The Tinkerings of Robert Noyce.” Esquire (December 1983): 346– 374.

Wood, Lamont。“发明个人电脑的人。” 《美国发明与技术遗产》（1994 年秋季）：第 64 页。

Wood, Lamont. “The Man Who Invented the PC.” American Heritage of Invention & Technology (fall 1994): 64.

Woodbury, William。“603-405 计算机。”第二届计算机研讨会论文集；1949 年 9 月。剑桥：哈佛大学出版社，1951 年。

Woodbury, William. “The 603-405 Computer.” Proceedings of a Second Symposium on Calculating Machinery; Sept. 1949. Cambridge: Harvard University Press, 1951.

Wozniak, Steve。“Apple II。” Byte（1977 年 5 月）：34–43。

Wozniak, Steve. “The Apple II.” Byte (May 1977): 34–43.

Wozniak, Steve。“访谈。” Byte（1985 年 1 月）：167–180。

Wozniak, Steve. “Interview.” Byte (January 1985): 167–180.

Yates, JoAnne。《通过沟通实现控制：美国管理体系的兴起》。巴尔的摩：约翰霍普金斯大学出版社，1989 年。

Yates, JoAnne. Control Through Communication: The Rise of System in American Management. Baltimore: Johns Hopkins University Press, 1989.

Zachary, G. Pascal。《精彩绝伦！微软为打造 Windows NT 和下一代产品而展开的激烈竞争》。纽约：Free Press，1994 年。

Zachary, G. Pascal. Showstopper! The Breakneck Race to Create Windows NT and the Next Generation at Microsoft. New York: Free Press, 1994.

祖博夫，肖莎娜。在智能机器时代。纽约：Basic Books，1988 年。Zuse，Konrad。计划施肥设备。波恩：数学与数据处理协会。 Zuse 档案，文件夹 #101/024，1944 年。

Zuboff, Shoshanna. In the Age of the Smart Machine. New York: Basic Books, 1988. Zuse, Konrad. Planfertigungsgeräte. Bonn: Gesellschaft für Mathematik und Datenverarbeitung. Zuse Archive, Folder #101/024, 1944.

楚泽，康拉德。 “程序员。” Zeitschrift für Angewandte Mathematik und Mechanik 32 (1952): 246。

Zuse, Konrad. “Der Programmator.” Zeitschrift für Angewandte Mathematik und Mechanik 32 (1952): 246.

指数

Index

1984，乔治·奥威尔的小说，274

1984, novel by George Orwell, 274

2001：太空漫游，斯坦利·库布里克执导，93–94，308

2001: A Space Odyssey, film by Stanley Kubrick, 93–94, 308

阿伯丁试验场。参见美国陆军弹道研究实验室

Aberdeen Proving Ground. See U.S. Army, Ballistic Research Laboratory

阿伯丁继电器计算器，18.另请参阅IBM，可插拔顺序继电器计算器

Aberdeen Relay Calculator, 18. See also IBM, Pluggable Sequence Relay Calculator

计算机处理器中的累加器，59–60。另请参阅计算机架构

Accumulator, in computer processor, 59–60. See also Architecture, computer

寻址，计算机，61，62，132。另请参阅架构，计算机；间接寻址；模式

Addressing, computer, 61, 62, 132. See also Architecture, computer; Indirect addressing; Modes

ADP。参见“自动数据处理冒险”，电脑游戏，210

ADP. See Automatic Data Processing Adventure, computer game, 210

AESOP（高级在线规划实验系统），252–253

AESOP (Advanced Experimental System for On-line Planning), 252–253

艾肯，Howard H.，13，51，81–82，212

Aiken, Howard H., 13, 51, 81–82, 212

空军，美国，31–31，51，55，65，71，140，179。另请参阅民兵弹道导弹和

Air Force, U.S., 31–31, 51, 55, 65, 71, 140, 179. See also Minuteman ballistic missile and

集成电路的发展，187

development of integrated circuit, 187

需要可靠的电路，179–182

need for reliable circuits, 179–182

赖特-帕特森空军基地，32，55

Wright-Patterson AFB, 32, 55

空军物资司令部，32

Air Material Command, 32

Aldus 公司，276

Aldus Corporation, 276

ALGOL 编程语言, 94, 98

ALGOL programming language, 94, 98

ALGOL-68 编程语言，101，106–107

ALGOL-68 programming language, 101, 106–107

算法，103

Algorithm, 103

艾伦，保罗，233–235，237

Allen, Paul, 233–235, 237

ALOHAnet，293

ALOHAnet, 293

Altair 8800 计算机，225–231

Altair 8800 computer, 225–231

阿姆达尔公司，164

Amdahl Corporation, 164

Amdahl，Gene，148，156，249–250

Amdahl, Gene, 148, 156, 249–250

美国管理系统 (AMS), 168

American Management Systems (AMS), 168

美国博彩公司，26

American Totalisator Company, 26

安德森，哈兰，127

Anderson, Harlan, 127

马克·安德森 303

Andreessen, Marc, 303

APL（一种编程语言），233，235，248

APL (A Programming Language), 233, 235, 248

阿波罗制导计算机，188–190

Apollo Guidance Computer, 188–190

阿波罗计划，124，188–189

Apollo, Project, 124, 188–189

Apollo（工作站制造商），281，285

Apollo (workstation manufacturer), 281, 285

苹果电脑公司，264

Apple Computer Company, 264

AppleTalk，275

AppleTalk, 275

Apple II 电脑，264–268，311

Apple II computer, 264–268, 311

磁盘驱动器开发，266

disk drive development, 266

Lisa 电脑，273–275

Lisa computer, 273–275

Macintosh 电脑，273–276，294

Macintosh computer, 273–276, 294

计算有限元分析的应用，147

Applications of computing finite element analysis, 147

工资单，33-34

payrolls, 33–34

对于个人电脑，230

for personal computers, 230

排版，109

typesetting, 109

风洞数据缩减，117

wind tunnel data reduction, 117

计算机体系结构，57–64。另请参阅累加器；寻址；索引寄存器；中断能力；微编程；冯·诺依曼体系结构；字长

Architecture, computer, 57–64. See also Accumulator; Addressing; Index registers; Interrupt capability; Microprogramming; von Neumann architecture; Word length

小型计算机，125

Minicomputer, 125

平行，21，24，196–197

Parallel, 21, 24, 196–197

阿姆斯特朗，兰斯洛特，44岁

Armstrong, Lancelot, 44

美国陆军，56

Army, U.S., 56

弹道研究实验室，马里兰州阿伯丁，18-19

Ballistic Research Laboratory, Aberdeen, Maryland, 18–19

通信兵团，180，224

Signal Corps, 180, 224

ARPA（高级研究计划局），117，140，194，259-261

ARPA (Advanced Research Projects Agency), 117, 140, 194, 259–261

IPTO（信息处理技术办公室），260 和 PDP-10，286

IPTO (Information Processing Techniques Office), 260 and PDP-10, 286

支持互联网，295

support for internet, 295

阿帕网

ARPANET

最初的动力，296

initial impetus for, 296

可用程序，298

programs available on, 298

在施乐-PARC，291–292

at Xerox-PARC, 291–292

安达信会计师事务所，32

Arthur Anderson & Co., 32

人工智能（AI），10

Artificial Intelligence (AI), 10

计算机编程艺术，103

Art of Computer Programming, the, 103

阿特维克，布鲁斯，278

Artwick, Bruce, 278

ASCC参见哈佛 Mark I

A.S.C.C. See Harvard Mark I

ASCII（美国信息交换标准代码）。另请参阅EBCDIC

ASCII (American Standard Code for Information Interchange). See also EBCDIC

IBM 和，193–194，268

IBM and, 193–194, 268

小型计算机和 152, 226, 251

minicomputers and, 152, 226, 251

起源，133

origins, 133

Assabet Mills（DEC 总部），127、129、137–139、141、312

Assabet Mills (DEC headquarters), 127, 129, 137–139, 141, 312

汇编程序，85、87–88

Assemblers, 85, 87–88

计算机协会（ACM），102

Association for Computing Machinery (ACM), 102

AT&T，65，66，172，282。另请参阅贝尔实验室

AT&T, 65, 66, 172, 282. See also Bell Laboratories

阿塔纳索夫，JV，38岁

Atanasoff, J. V., 38

Atlas 计算机 (ERA)。参见工程研究协会，1101

Atlas computer (ERA). See Engineering Research Associates, 1101

Atlas 计算机（曼彻斯特）。参见Ferranti、Atlas

Atlas computer (Manchester). See Ferranti, Atlas

阿特拉斯导弹，67

Atlas missile, 67

原子能委员会，30

Atomic Energy Commission, 30

加拿大原子能，128–129

Atomic Energy of Canada, 128–129

AUTOCODE 编程语言，113

AUTOCODE programming language, 113

自动数据处理（ADP）, 168

Automatic Data Processing (ADP), 168

自动工资单，168

Automatic payrolls, 168

自动编程。参见编译器自动柜员机（ATM），80–81

Automatic programming. See Compilers Automatic teller machine (ATM), 80–81

自动化，32

Automation, 32

Autonetics。参见北美航空自动装配，180

Autonetics. See North American Aviation Auto-Sembly, 180

B 编程语言，106

B programming language, 106

巴科斯范式 (BNF), 95

Backus-Naur Form (BNF), 95

巴克斯，约翰，79，91

Backus, John, 79, 91

弹道导弹预警系统（BMEWS），71

Ballistic Missile Early Warning System (BMEWS), 71

弹道研究实验室，马里兰州阿伯丁，18-19

Ballistic Research Laboratory, Aberdeen, Md., 18–19

美国银行，55

Bank of America, 55

用于 Altair 的 BASIC 编程语言，232–236

BASIC programming language for the Altair, 232–236

Apple II 电脑，264–265

for the Apple II, 264–265

达特茅斯学院，203–205

at Dartmouth, 203–205

Microsoft BASIC。请参阅Microsoft 小型计算机改编版，205–206，235–236

Microsoft BASIC. See Microsoft minicomputer adaptations of, 205–206, 235–236

只读存储器 (ROM)，240，264–265

on a read-only memory (ROM), 240, 264–265

批处理，74，77–78，122–123

Batch processing, 74, 77–78, 122–123

BBN（Bolt Beranek 和 Newman），128，259

BBN (Bolt Beranek and Newman), 128, 259

安迪·贝克托尔斯海姆 282

Bechtolsheim, Andy, 282

贝尔实验室，64–65、155–156、203、282

Bell Laboratories, 64–65, 155–156, 203, 282

V 型电脑，18

Model V computer, 18

VI 型计算机，149

Model VI computer, 149

TRADIC 计算机，65

TRADIC computer, 65

贝尔，C.戈登，198–199，211，246

Bell, C. Gordon, 198–199, 211, 246

本迪克斯，42–43，165

Bendix, 42–43, 165

G-15 计算机，42–43，132

G-15 computer, 42–43, 132

贝内特，爱德华，253

Bennett, Edward, 253

伯克利计算机公司（BCC），259

Berkeley Computer Corporation (BCC), 259

伯克利 UNIX。请参阅UNIX Berkeley，Edmund C.，27

Berkeley UNIX. See UNIX Berkeley, Edmund C., 27

加州大学伯克利分校，140，259

Berkeley, University of California at, 140, 259

伯纳斯-李，蒂姆，302

Berners-Lee, Tim, 302

艾瑞克·比娜，303

Bina, Eric, 303

BINAC 计算机，23

BINAC computer, 23

BIOS（基本输入/输出系统），239，271，277

BIOS (Basic Input/Output System), 239, 271, 277

位图显示，262

Bit-mapped display, 262

BITNET，299

BITNET, 299

埃里克·布洛赫（Eric Bloch） 51 岁

Bloch, Eric, 51

Boehm, Barry, 82, 170

博格斯，大卫，291–292

Boggs, David, 291–292

博纳维尔电力管理局，235

Bonneville Power Administration, 235

Bowmar Brain 计算器，213

Bowmar Brain calculator, 213

查尔斯·布拉德肖，167

Bradshaw, Charles, 167

布拉姆霍尔，马克，236

Bramhall, Mark, 236

布兰德、斯图尔特、207、221、258、260、280

Brand, Stewart, 207, 221, 258, 260, 280

丹尼尔·布里克林 267

Bricklin, Daniel, 267

布鲁克斯，弗雷德，101，148，156，164

Brooks, Fred, 101, 148, 156, 164

浏览器，超文本软件，302

Browser, software for hypertext, 302

布赫霍尔茨，沃纳，151

Buchholz, Werner, 151

Bull, Machines（法国计算机公司），172

Bull, Machines (French computer company), 172

个人电脑的公告板系统（BBS），298

Bulletin Board Systems (BBS), for personal computers, 298

BUNCH（Burroughs、UNIVAC、NCR、Control Data、Honeywell），143、171–173、248

BUNCH (Burroughs, UNIVAC, NCR, Control Data, Honeywell), 143, 171– 173, 248

伯克斯·亚瑟 44 岁

Burks, Arthur, 44

宝来公司，41，51，66–67，213

Burroughs Corporation, 41, 51, 66–67, 213

加法机和计算器，119

adding machines and calculators, 119

E-101 电脑, 67

E-101 computer, 67

总线、计算机、199、228–229

Bus, computer, 199, 228–229

布什，万尼瓦尔，301

Bush, Vannevar, 301

Busicom，218，220–221

Busicom, 218, 220–221

字节杂志，310

Byte magazine, 310

字节（位分组），151、193–194

Byte (grouping of bits), 151, 193–194

C-Cubed（计算机中心公司），211，235

C-Cubed (Computer Center Corporation), 211, 235

C++ 编程语言，79

C++ programming language, 79

C 编程语言，91，106，283

C programming language, 91, 106, 283

CADAC 计算机，40。另请参阅计算机研究公司

CADAC computer, 40. See also Computer Research Corporation

CADET。参见IBM 1620 计算机

CADET. See IBM 1620 computer

CalComp，164

CalComp, 164

计算器，可编程袖珍型，214–216。另请参阅制造商下的个别品牌

Calculators, programmable pocket, 214–216. See also individual brands under manufacturers

西奥多·卡洛（Theodore Callow），33岁

Callow, Theodore, 33

卡尼翁，杆，277

Canion, Rod, 277

爱德华·坎农 13 岁

Cannon, Edward, 13

卡内基梅隆大学，259

Carnegie-Mellon University, 259

卡特，吉米，255–256

Carter, Jimmy, 255–256

卡西欧，计算器制造商，213

Casio, calculator manufacturer, 213

CCC。参见计算机控制公司

CCC. See Computer Controls Corporation

CDC。参见控制数据公司人口普查局，美国，第 26-27、29、31 页

CDC. See Control Data Corporation Census Bureau, U.S., 26–27, 29, 31

Centrallab。参见Globe Union CERN，第 301–302 页

Centrallab. See Globe Union CERN, 301–302

通道，输入/输出，62–63、128。

Channels, input/output, 62–63, 128.

另请参阅建筑、计算机 Charactron、显示管、38

See also Architecture, computer Charactron, display tube, 38

CICS（客户信息控制系统），106，201

CICS (Customer Information Control System), 106, 201

纽约城市大学，299

City University of New York, 299

克拉克，吉姆，303

Clark, Jim, 303

克拉克，韦斯，130

Clark, Wes, 130

微电子组装洁净室，181

Clean rooms, for microelectronics assembly, 181

克隆。请参阅兼容性 COBOL 编程语言，91–93、113、163

Clones. See Compatibility COBOL programming language, 91– 93, 113, 163

科克，约翰，288，290

Cocke, John, 288, 290

冷战，7–8、11、140、169–170

Cold War, 7–8, 11, 140, 169–170

科尔顿制造公司, 53

Colton Manufacturing Company, 53

哥伦比亚大学，18

Columbia University, 18

商业信贷，172

Commercial Credit, 172

Commodore，PET 计算机，264

Commodore, PET computer, 264

康柏，277

Compaq, 277

兼容性，计算机。另请参阅与 Altair 兼容的插头制造商，231，239–240

Compatibility, computer. See also Plug- compatible manufacturers with Altair, 231, 239–240

IBM 大型机，161–163

IBM mainframe, 161–163

IBM PC（“克隆”），277–278，294

IBM PC (“clones”), 277–278, 294

编译器，85，A-0，A-1，A-2，85

Compilers, 85, A-0, A-1, A-2, 85

FLOW-MATIC、MATH-MATIC、92–93

FLOW-MATIC, MATH-MATIC, 92–93

计算计，47–48

Comptometer, 47–48

计算机控制公司（CCC），194

Computer Controls Corporation (CCC), 194

计算机租赁公司，159

Computer leasing companies, 159

计算机图书馆/梦想机器，216，301

Computer Lib/Dream Machines, 216, 301

计算机内存。参见磁芯；磁带；RAM

Computer memory. See Magnetic core; Magnetic tape; RAM

计算机网络，120。另请参阅

Computer networking, 120. See also

ARPANET；以太网；IBM 系统

ARPANET; Ethernet; IBM Systems

网络架构；互联网

Network Architecture; Internet

计算机研究公司（CRC），38，40，66

Computer Research Corporation (CRC), 38, 40, 66

计算机科学，101–103，201–203

Computer science, 101–103, 201–203

计算机科学公司（CSC），92，168

Computer Sciences Corporation (CSC), 92, 168

计算机终端公司，221

Computer Terminal Corporation, 221

计算机使用公司，167

Computer Usage Company, 167

《计算机世界》杂志，121

Computerworld Magazine, 121

综合工程，41，67

Consolidated Engineering, 41, 67

控制数据公司 (CDC)，43，125–126，161，172。另请参阅Norris，

Control Data Corporation (CDC), 43, 125–126, 161, 172. See also Norris,

威廉；克雷，西摩

William; Cray, Seymour

CDC 160, 160A 计算机, 59, 126, 129, 132

CDC 160, 160A computers, 59, 126, 129, 132

CDC 1604 计算机，126

CDC 1604 computer, 126

CDC 6600 计算机，161

CDC 6600 computer, 161

服务局，167

service bureau, 167

康维尔（联合沃尔蒂飞机公司），38

Convair (Consolidated Vultee Aircraft Corporation), 38

对话监控系统 (CMS)，157，200

Conversational Monitoring System (CMS), 157, 200

核心转储，97，99–100

Core dump, 97, 99–100

磁芯存储器。请参阅磁芯存储器

Core memory. See Magnetic core memory

CP/M-86 操作系统，268

CP/M-86 operating system, 268

CP/M 操作系统，232、238–240、262、270 和 Microsoft，265

CP/M operating system, 232, 238–240, 262, 270 and Microsoft, 265

CPT，255

CPT, 255

克拉布特里，鲍勃，124

Crabtree, Bob, 124

克劳福德，佩里·O.，38岁

Crawford, Perry O., 38

克雷研究公司，173

Cray Research, 173

Cray，Seymour，37，125–126，161，172–173

Cray, Seymour, 37, 125–126, 161, 172–173

CRC。参见计算机研究公司 Cromemco 个人计算机，270

CRC. See Computer Research Corporation Cromemco personal computer, 270

交叉编译，223，238

Cross-compiling, 223, 238

CSAW（通信补充活动——华盛顿），36

CSAW (Communications Supplementary Activity— Washington), 36

CTSS（兼容分时系统），155–156，203，208

CTSS (Compatible Time-Sharing System), 155–156, 203, 208

课程 '68，102-103

Curriculum ’68, 102–103

DARPA（国防高级研究计划局）。请参阅ARPA

DARPA (Defense Advanced Research Projects Agency). See ARPA

达特茅斯学院，55，70分时，200，250

Dartmouth College, 55, 70 time sharing at, 200, 250

DASD（直接访问存储设备），200

DASD(Direct Access Storage Device), 200

Data General Corporation，195，246，305–306。另请参阅《新机器的灵魂》，

Data General Corporation, 195, 246, 305–306. See also Soul of a New Machine, The

施乐帕洛阿尔托研究中心的计算机，291

computers at Xerox-PARC, 291

对其他计算机设计的影响，265，305

influence on designs of other computers, 265, 305

新星，195，199，227，273

Nova, 195, 199, 227, 273

超新星，195-197

Super Nova, 195–197

Datamation杂志，68

Datamation magazine, 68

Datapoint 公司，221 2200 终端，251

Datapoint Corporation, 221 2200 terminal, 251

数据处理，定义，48

Data processing, definition, 48

大卫杜夫，蒙特，235

Davidoff, Monte, 235

dBase III 数据库程序, 293

dBase III database program, 293

DEC.参见数字设备公司

DEC. See Digital Equipment Corporation

德卡斯特罗，埃德森，130，195

DeCastro, Edson, 130, 195

国防计算器。参见IBM 701computer

Defense calculator. See IBM 701computer

美国国防部，计算支持，91，112，258，289。另请参阅个别服务

Defense Department, U.S., support for computing, 91, 112, 258, 289. See also individual services

民主与计算，9

Democracy, and computing, 9

约翰·迪博尔德 32 岁

Diebold, John, 32

数字化工作，140

Digital at Work, 140

数字设备公司

Digital Equipment Corporation

（DEC），88，101，110，128，159，191，243。另请参阅Assabet Mills；Bell，C. Gordon；Olsen，Kenneth

(DEC), 88, 101, 110, 128, 159, 191, 243. See also Assabet Mills; Bell, C. Gordon; Olsen, Kenneth

和 BASIC，236 文化，136–139

and BASIC, 236 culture, 136–139

DECUS（用户组），88

DECUS (users group), 88

财务损失，305–306

financial losses, 305–306

创始，127

founding, 127

操作系统，238–240

operating systems, 238–240

收入增长，136–137

revenue growth, 136–137

和 UNIX，247，284

and UNIX, 247, 284

数字设备公司计算机

Digital Equipment Corporation computers

DECsystem 10、DECsystem 20。参见PDP-10

DECsystem 10, DECsystem 20. See PDP-10

PDP-1、127–128、204、255

PDP-1, 127–128, 204, 255

PDP-5, 130

PDP-6, 139, 208, 209

PDP-7、106、157、283

PDP-7, 106, 157, 283

PDP-8、110、129–137、194、218、228

PDP-8, 110, 129–137, 194, 218, 228

PDP 8A, 135, 244

PDP-10、139、157、166、208、210、215、235、286

PDP-10, 139, 157, 166, 208, 210, 215, 235, 286

PDP-11、157、198–200、204、205、243–244、275、283、287

PDP-11, 157, 198–200, 204, 205, 243– 244, 275, 283, 287

PDP-11/20, 205

PDP-11/45, 205

PDP-X，195

PDP-X, 195

外围设备, 132, 208, 212, 247, 251, 286

peripheral equipment, 132, 208, 212, 247, 251, 286

个人电脑，245，287

personal computers, 245, 287

彩虹，287

Rainbow, 287

VAX、101、243–247、284–287

VAX, 101, 243–247, 284–287

数字研究，270

Digital Research, 270

Dijksterhuis，EJ，308

Dijksterhuis, E. J., 308

迪克斯特拉，埃兹格，104

Dijkstra, Edsger, 104

DIP（双列直插式封装），192

DIP (Dual In-line Package), 192

直接内存访问，93.另请参阅

Direct memory access, 93. See also

建筑学、计算机

Architecture, computer

磁盘操作系统。请参阅DOS、MS-DOS

Disk Operating System. See DOS, MS-DOS

磁盘存储，69。另请参阅软盘

Disk storage, 69. See also Floppy disk

域操作系统, 281

Domain operating system, 281

多里奥特，乔治，127，140

Doriot, George, 127, 140

DOS（磁盘操作系统），术语的起源，237

DOS (Disk Operating System), origin of term, 237

Dobb 博士的计算机健美操和正畸学杂志，238–239

Dr. Dobb’s Journal of Computer Calisthenics and Orthodontia, 238–239

Draper 实验室。参见麻省理工学院仪器实验室 Draper，Charles Stark，工程奖，79

Draper Labs. See MIT Instrumentation Laboratory Draper, Charles Stark, Engineering Prize, 79

哑终端，251

Dumb terminals, 251

转储。请参阅核心转储

Dump. See Core dump

Dynaco 套件，229

Dynaco kits, 229

动态地址转换，155.另请参阅计算机体系结构

Dynamic address translation, 155. See also Architecture, computer

动态 RAM (DRAM)。参见RAM

Dynamic RAM (DRAM). See RAM

EBCDIC（扩展二进制编码

EBCDIC (Extended Binary Coded

十进制交换码）、152、193、251

Decimal Interchange Code), 152, 193, 251

Eckert, J. Presper, 13, 18, 22, 25–27, 69, 179。另请参阅

Eckert, J. Presper, 13, 18, 22, 25–27, 69, 179. See also

埃克特-莫奇利计算机公司

Eckert-Mauchly Computer Corporation

埃克特，华莱士，18岁

Eckert, Wallace, 18

Eckert-Mauchly 计算机公司，13–16、22–23、25–27、45、89。另请参阅雷明顿·兰德

Eckert-Mauchly Computer Corporation, 13–16, 22–23, 25–27, 45, 89. See also Remington Rand

大卫·埃德尔斯坦，249

Edelstein, David, 249

EDSAC，23，84

EDSAC, 23, 84

EDVAC，18，21，25，89

EDVAC, 18, 21, 25, 89

艾森豪威尔，德怀特·D·艾森豪威尔，31–32，258

Eisenhower, Dwight D., 31–32, 258

艾斯勒，保罗，180

Eisler, Paul, 180

电数据，67

Electrodata, 67

电子计算机公司，41

Electronic Computer Corporation, 41

电子数据系统 (EDS), 169

Electronic Data Systems (EDS), 169

电子前沿基金会，312

Electronic Frontier Foundation, 312

电子邮件，起源于 ARPANET，298

Electronic mail, origins on ARPANET, 298

电子多元化，135

Electronics Diversified, 135

电子杂志，309

Electronics magazine, 309

艾尔金德，杰罗姆，259

Elkind, Jerome, 259

一台计算机对另一台计算机的模拟，149，151

Emulation, of one computer by another, 149, 151

Engelbart, Doug, 207, 259–260。另请参阅鼠标、NLS 和超文本，301, 303

Engelbart, Doug, 207, 259–260. See also Mouse, NLS and hypertext, 301, 303

工程研究协会（ERA），36–39, 45 1101

Engineering Research Associates (ERA), 36–39, 45 1101

计算机，37 1103 计算机，38

computer, 37 1103 computer, 38

英语，比尔，260

English, Bill, 260

霍华德·恩格斯特罗姆 36 岁

Engstrom, Howard, 36

ENIAC，7、15、18、20-21

ENIAC, 7, 15, 18, 20–21

EPROM（可擦除可编程只读存储器），220

EPROM (erasable programmable read only memory), 220

ERA。参见工程研究协会

ERA. See Engineering Research Associates

ERMA（电子会计方法），55

ERMA (Electronic Method of Accounting), 55

以太网，261，263 和互联网，297

Ethernet, 261, 263 and Internet, 297

发明，291–292

invention of, 291–292

向 DEC、Intel 和 Xerox 授予许可，292

license to DEC, Intel, and Xerox, 292

埃文斯，鲍勃·O.，247

Evans, Bob O., 247

埃文斯，大卫，285

Evans, David, 285

埃弗里特，罗伯特，140

Everett, Robert, 140

尤因，格伦，240

Ewing, Glen, 240

“昂贵的打字机”，255

“Expensive typewriter,” 255

电子电路制造厂（制造工厂），181

Fabs (fabrication plants), for electronic circuits, 181

FACT 编译器, 92, 168

FACT compiler, 92, 168

费德里科·费金 220

Faggin, Federico, 220

费尔柴尔德相机和仪器，186

Fairchild Camera and Instrument, 186

飞兆半导体，179，185–188，196

Fairchild Semiconductor, 179, 185– 188, 196

SYMBOL 计算机，193

SYMBOL computer, 193

费尔柴尔德，谢尔曼，186

Fairchild, Sherman, 186

费尔克拉夫，约翰，148

Fairclough, John, 148

美国联邦航空管理局，165

Federal Aviation Administration, 165

苏黎世联邦理工大学（ETH），83-84

Federal Technical University (ETH), Zurich, 83–84

费尔森斯坦，李，226

Felsenstein, Lee, 226

费尔特和塔兰特，47

Felt and Tarrant, 47

弗格森，尤金，186

Ferguson, Eugene, 186

Ferranti 有限公司，245

Ferranti, Ltd., 245

Atlas 计算机，245–246

Atlas computer, 245–246

第五代，日本计算计划，11

Fifth generation, Japanese computing initiative, 11

文件服务器，术语的起源，294

File Server, origin of term, 294

“EDVAC 报告初稿”，第 21-22 页。另请参阅约翰·冯·诺依曼

“First Draft of a Report on the EDVAC,” 21–22. See also von Neumann, John

柔版印刷机，42，133

Flexowriter, 42, 133

飞行模拟器，程序，278

Flight Simulator, program, 278

倒装芯片模块，179–180

Flip-Chip modules, 179–180

浮点运算，63–64，213

Floating point arithmetic, 63–64, 213

软盘存储，158，231–232

Floppy disk storage, 158, 231–232

Apple II 电脑，266–267

for the Apple II, 266–267

对于个人电脑，236

for personal computers, 236

FLOW-MATIC。参见编译器

FLOW-MATIC. See Compilers

福特汽车公司，32，66

Ford Motor Company, 32, 66

亨利·福特，214

Ford, Henry, 214

罗伯特·福雷斯特 (Robert Forest) 68 岁

Forest, Robert, 68

福雷斯特，杰伊，140

Forrester, Jay, 140

福赛斯，乔治，102

Forsythe, George, 102

FORTRAN，64，69，79，90–91，113，200

FORTRAN, 64, 69, 79, 90–91, 113, 200

FORTRAN 监控系统，100

FORTRAN Monitor System, 100

对于个人电脑，232–233

for personal computers, 232–233

弗兰克斯顿，罗伯特，267

Frankston, Robert, 267

弗雷德金，爱德华，129

Fredkin, Edward, 129

Friden 计算器，119，212

Friden calculators, 119, 212

弗罗曼·多夫，220

Frohman, Dov, 220

FTP（文件传输协议），298

FTP (file transfer protocol), 298

富士通，165

Fujitsu, 165

富布赖特，J.威廉，258

Fulbright, J. William, 258

加尔布雷斯，约翰·肯尼斯，141

Galbraith, John Kenneth, 141

加勒，伯尼，96，104

Galler, Bernie, 96, 104

加德纳-丹佛公司，133

Gardner-Denver Corporation, 133

Gates, William III (Bill), 211, 233–237, 270。另请参阅BASIC；Microsoft Corporation；MS-DOS

Gates, William III (Bill), 211, 233–237, 270. See also BASIC; Microsoft Corporation; MS-DOS

具有 PDP-10、211、235、286 的使用经验

experience with PDP-10, 211, 235, 286

美国审计总署（GAO），121

General Accounting Office (GAO), 121

通用电气公司，54–55，157–158

General Electric Corporation, 54–55, 157–158

作为 UNIVAC 的客户，30、32–33

as customer for UNIVAC, 30, 32–33

GE-Calma，281

GE-Calma, 281

GE 635 计算机，155

GE 635 computer, 155

OARAC，55岁

OARAC, 55

分时系统，155，203–204，235，250

time-sharing system, 155, 203–204, 235, 250

通用磨坊公司，53

General Mills Corporation, 53

计算机的世代，6

Generations, of computers, 6

锗，电路材料，183

Germanium, material for circuit, 183

玻璃 TTY（玻璃电传打字机），251

Glass TTY (Glass Teletype), 251

格伦，约翰，123

Glenn, John, 123

环球联合公司，180，182，193

Globe-Union Company, 180, 182, 193

繁荣时期（约 1966-1968 年），159，199 “Go-To 语句考虑

Go-go years (ca. 1966–1968), 159, 199 “Go-To Statement Considered

有害”，104

Harmful,” 104

戈德曼，雅各布，258

Goldman, Jacob, 258

赫尔曼·戈德斯坦（Herman Goldstine）21岁，44岁

Goldstine, Herman, 21, 44

Gopher，互联网软件，299–300

Gopher, software for the Internet, 299–300

格雷厄姆，J.韦斯利，202–203

Graham, J. Wesley, 202–203

图形用户界面 (GUI)。请参阅

Graphical User Interface (GUI). See

GEM、顶视图、Microsoft Windows、X-Windows

GEM, Top View, Microsoft Windows, X-Windows

大社会，计算需求，110

Great Society, computing needs for, 110

格罗施，赫伯特，177–178

Grosch, Herbert, 177–178

格罗施定律，177–178，263，291

Grosch’s Law, 177–178, 263, 291

格罗夫，安德鲁，198，297

Grove, Andrew, 198, 297

格罗夫定律，297

Grove’s Law, 297

格尔利，本，127

Gurley, Ben, 127

哈达德·杰瑞尔（Haddad, Jerrier） 34 岁

Haddad, Jerrier, 34

HAL，2001 年的虚构计算机：

HAL, fictional computer in 2001: A

太空漫游，93–94，208

Space Odyssey, 93–94, 208

霍尔，埃尔登，188，190

Hall, Eldon, 188, 190

哈里斯，吉姆，277

Harris, Jim, 277

哈佛大学，236

Harvard University, 236

商学院，127，267

Business School, 127, 267

Mark I 计算器，13，81–82，199

Mark I Calculator, 13, 81–82, 199

Mark III 计算器，82–83

Mark III Calculator, 82–83

Mark IV 计算器, 51

Mark IV Calculator, 51

HASP（休斯顿自动后台处理优先级），124

HASP (Houston Automatic Spooling Priority), 124

希思基特，229–230

Heathkits, 229–230

亨德里·加德纳，194

Hendrie, Gardner, 194

约翰·轩尼诗，289–290

Hennessy, John, 289–290

惠普公司，211

Hewlett-Packard Corporation, 211

HP-35 计算器，213

HP-35 calculator, 213

HP-65 计算器，189，213–215，226

HP-65 calculator, 189, 213–215, 226

HP-2000小型机，205

HP-2000 minicomputer, 205

HP-9000工作站，281

HP-9000 workstation, 281

HP-9100A 计算器，212

HP-9100A calculator, 212

Hi-Rel（高可靠性）。参见Minuteman

Hi-Rel (High Reliability). See Minuteman

赫尔穆特·霍尔策尔（Helmut Hoeltzer），167

Hoeltzer, Helmut, 167

Hoerni，Jean，186

Hoerni, Jean, 186

霍夫，马西安·E.（特德），218–221

Hoff, Marcian E. (Ted), 218–221

霍尔伯顿，弗朗西斯·E.，89–90

Holberton, Frances E., 89–90

赫尔曼·霍勒里斯（Herman Hollerith） 26 岁

Hollerith, Herman, 26

自制计算机俱乐部，216，264

Homebrew Computer Club, 216, 264

霍尼韦尔公司, 54, 56, 157, 172, 194

Honeywell Corporation, 54, 56, 157, 172, 194

DDP-516, 194–195

H-200, 119, 151

H-800, 116

分时系统，204

time sharing systems, 204

霍珀，格雷斯·默里，27，29，81–82，85，93–94

Hopper, Grace Murray, 27, 29, 81–82, 85, 93–94

房子，查克，211

House, Chuck, 211

HTML（超文本标记语言），302

HTML (hypertext markup language), 302

HTTP（超文本传输协议），302

HTTP (hypertext transfer protocol), 302

哈斯基，哈利，42-43

Huskey, Harry, 42–43

海亚特，吉尔，218，220

Hyatt, Gil, 218, 220

Hypercard，适用于 Macintosh 的软件，302

Hypercard, software for Macintosh, 302

超文本，起源，301

Hypertext, origins, 301

IAS 计算机。请参阅高等研究院

IAS computer. See Institute for Advanced Study

IBM 公司，8，14，67-69

IBM Corporation, 8, 14, 67–69

反垄断诉讼，160，164，170-171，248-250

antitrust suits against, 160, 164, 170– 171, 248–250

租赁政策，124，128，159

leasing policies, 124, 128, 159

磁带开发，36

magnetic tape development, 36

市场份额、收入、36、110–111、143、145、256、304–305

market share, revenues, 36, 110–111, 143, 145, 256, 304–305

研究实验室，160

research laboratories, 160

为商业和科学设立单独的机器生产线，43

separate lines of machine for business and science, 43

作为其他制造商的 I/O 设备供应商，40–41

as supplier of I/O equipment to other manufacturers, 40–41

System/360 系列，64、100、113、119、144、157、160

System/360 series, 64, 100, 113, 119, 144, 157, 160

公告，144

announcement, 144

建筑，145–149，157，287

architecture, 145–149, 157, 287

电路，183，190–191

circuits, 183, 190–191

型号 44, 248

model 44, 248

型号 50、117、150

model 50, 117, 150

型号 65, 146

model 65, 146

型号 67、117、156–157

model 67, 117, 156–157

型号 75、124、202–203

model 75, 124, 202–203

型号 91、162

model 91, 162

System/370 系列，200、205、232

System/370 series, 200, 205, 232

计划更换，161，252

plans to replace, 161, 252

虚拟内存输入, 246

virtual memory in, 246

Thomas J. Watson 计算局，18

Thomas J. Watson Computing Bureau, 18

令牌环网络, 292

Token Ring networking, 292

拆分决定，105，168

unbundling decision, 105, 168

和 UNIVAC，34

and UNIVAC, 34

IBM 计算机和制表机。（另请参阅System/360 系列；System/370

IBM computers and tabulators. (See also System/360 series; System/370

系列）601、603、604、605

series) 601, 603, 604, 605

倍增拳，19 604

multiplying punches, 19 604

乘数，43 650

multiplier, 43 650

计算机, 43–44, 70, 76, 101, 113, 119

computer, 43–44, 70, 76, 101, 113, 119

701 计算机, 30, 34–36, 38, 45

701 computer, 30, 34–36, 38, 45

702 计算机, 36–37, 58

702 computer, 36–37, 58

704 计算机, 64, 68, 88, 91, 122

704 computer, 64, 68, 88, 91, 122

801电脑，288

801 computer, 288

1401 计算机, 59, 73–76, 116, 119, 132, 147, 151

1401 computer, 59, 73–76, 116, 119, 132, 147, 151

1620计算机，75–76，147，218

1620 computer, 75–76, 147, 218

3030 系列, 161

3030 series, 161

4300 系列，161

4300 series, 161

5100 个人电脑，248 7030（参见 STRETCH）

5100 personal computer, 248 7030 (See STRETCH)

7040电脑，116

7040 computer, 116

7070 计算机, 109, 112–113, 119, 147

7070 computer, 109, 112–113, 119, 147

7090 计算机, 71, 116, 123, 127, 147, 155

7090 computer, 71, 116, 123, 127, 147, 155

7094 计算机, 71–74, 116, 155

7094 computer, 71–74, 116, 155

9020电脑，150

9020 computer, 150

AS/400 系列，161、250、252

AS/400 series, 161, 250, 252

卡片编程计算器，18–20，22，45

Card Programmed Calculator, 18–20, 22, 45

直接耦合系统，116–118

Direct Couple System, 116–118

未来系统（FS），252

Future System (FS), 252

个人电脑（PC），217，268–269，272–275，294

Personal Computer (PC), 217, 268– 269, 272–275, 294

R/6000工作站，290

R/6000 workstation, 290

RT，288

RT, 288

斯坎普，233

SCAMP, 233

拉伸，151

STRETCH, 151

系统/3，158

System/3, 158

系统/23

System/23

数据管理员，269

Datamaster, 269

系统/38, 250

System/38, 250

XT，298

XT, 298

IBM 外设及其他设备

IBM peripherals and other equipment

027 键盘穿孔机, 109

027 keypunch, 109

029 键盘，145 350

029 keypunch, 145 350

磁盘存储单元，70，200

Disk Storage Unit, 70, 200

1403 链式打印机，77，147

1403 chain printer, 77, 147

3270 终端，251

3270 terminal, 251

MTST（磁带电动打字机），255

MTST (Magnetic Tape Selectric Typewriter), 255

电动打字机，146，153

Selectric Typewriter, 146, 153

ICL（国际计算机有限公司），249

ICL (International Computers, Ltd.), 249

IFIP（国际信息处理联合会），107

IFIP (International Federation of Information Processing), 107

伊利亚克四世（196–197）

Illiac-IV, 196–197

伊姆萨伊，231，240

IMSAI, 231, 240

医学索引, 250

Index Medicus, 250

索引寄存器，61，151。另请参阅体系结构，计算机

Index Registers, 61, 151. See also Architecture, computer

高等研究院（IAS），34，44

Institute for Advanced Study (IAS), 34, 44

集成电路（IC）

Integrated Circuit (IC)

首次用于计算机，163–166，191

first use in computers, 163–166, 191

发明，179，182–187

invention of, 179, 182–187

在小型计算机中，178

in minicomputers, 178

MSI、LSI、VLSI（中型、大型、超大规模集成）、195

MSI, LSI, VLSI (medium, large, very large scale integration), 195

英特尔公司

Intel Corporation

1103 内存芯片，197–198

1103 memory chip, 197–198

4004 微处理器，219–220

4004 microprocessor, 219–220

8008 微处理器，221–222

8008 microprocessor, 221–222

8080 微处理器，219，222，226，228，233，235

8080 microprocessor, 219, 222, 226, 228, 233, 235

8086 处理器，270，279

8086 processor, 270, 279

8088微处理器，277

8088 microprocessor, 277

80386 微处理器，294

80386 microprocessor, 294

创始，198，217–221

founding, 198, 217–221

Intellec 开发系统，222– 224

Intellec Development systems, 222– 224

界面管理器，图形

Interface Manager, graphical

界面，276。另请参阅Microsoft Corporation、Windows

interface, 276. See also Microsoft Corporation, Windows

接口消息处理器 (IMP)，

Interface message processors (IMP),

291.另请参阅霍尼韦尔 DDP-516

291. See also Honeywell DDP-516

美国国税局，109，118-122

Internal Revenue Service, 109, 118– 122

互联网，128。另请参阅ARPANET，

Internet, 128. See also ARPANET,

万维网的起源，295–297

World Wide Web origins, 295–297

互联网设备，需要，304

internet appliance, need for, 304

中断能力，38。另请参阅

Interrupt capability, 38. See also

建筑学、计算机

Architecture, computer

IPTO。参见ARPA

IPTO. See ARPA

伊斯哥尼斯，亚历克，135

Issigonis, Alec, 135

ITS（不兼容分时系统），208

ITS (Incompatible Time Sharing System), 208

ITT 公司，129

ITT Corporation, 129

艾弗森，肯尼斯，233，248

Iverson, Kenneth, 233, 248

日本，计算机发展，11.另请参阅

Japan, computer developments in, 11. See also

富士通喷气推进实验室，168

Fujitsu Jet Propulsion Laboratory, 168

乔布斯，史蒂文，264，274

Jobs, Steven, 264, 274

Johnniac 电脑，45

Johnniac computer, 45

琼斯，弗莱彻，168

Jones, Fletcher, 168

欢乐，95岁

JOVIAL, 95

Joy，Bill，282–284，289

Joy, Bill, 282–284, 289

美国司法部，164，169-170，248-250

Justice Department, U.S., 164, 169– 170, 248–250

卡尔，布鲁斯特，300

Kahle, Brewster, 300

堪萨斯城标准，用于数据存储，231

Kansas City standard, for data storage, 231

卡普尔，米奇，312

Kapor, Mitch, 312

卡岑巴赫，尼古拉斯，249

Katzenbach, Nicholas, 249

凯·艾伦，259

Kay, Alan, 259

Kemeney，John G.，203

Kemeney, John G., 203

肯巴克-1, 225

Kenbak-1, 225

肯尼迪，约翰·F·肯尼迪，110，113

Kennedy, John F., 110, 113

科尔萨，维诺德，282

Kholsa, Vinod, 282

基德，特蕾西，246，305

Kidder, Tracy, 246, 305

基尔比，杰克，179，182–185

Kilby, Jack, 179, 182–185

基尔达尔，加里，223，233，236–239，270–271

Kildall, Gary, 223, 233, 236–239, 270– 271

金·斯科特（Kim, Scott） 81 岁

Kim, Scott, 81

克努斯·唐纳德（Knuth, Donald）86、90、103

Knuth, Donald, 86, 90, 103

库尔茨，托马斯·E.，203

Kurtz, Thomas E., 203

兰普森，巴特勒，259

Lampson, Butler, 259

兰开斯特，唐，226–231

Lancaster, Don, 226–231

Lanier，文字处理公司，255–256

Lanier, word processing company, 255–256

兰宁，JH，86岁

Laning, J. H., 86

兰萨罗塔，桑托，68岁

Lanzarotta, Santo, 68

劳滕伯格，弗兰克，168

Lautenberg, Frank, 168

劳伦斯·利弗莫尔[加利福尼亚州]

Lawrence Livermore [California]

国家实验室，31，128，161

National Laboratory, 31, 128, 161

Lear-Siegler ADM 33“哑终端”，251，284

Lear-Siegler ADM 33 “dumb terminal,” 251, 284

利里，蒂莫西，309，311

Leary, Timothy, 309, 311

莱斯科，159

Leasco, 159

LEO（里昂电子办公室）计算机，10

LEO (Lyons Electronic Office) computer, 10

LGP-30。参见Librascope/General Precision

LGP-30. See Librascope/General Precision

解放者（霍尼韦尔 H-200 项目），151

Liberator (Honeywell program for H-200), 151

Librascope/General Precision，42

Librascope/General Precision, 42

LGP-30, 42, 101, 132

Licklider，JCR，259，306

Licklider, J. C. R., 259, 306

LINC，130，132，212

LINC, 130, 132, 212

线性规划，31

Linear Programming, 31

链接列表，90

Linked lists, 90

莱诺排字机，109

Linotype machine, 109

LISTSERV 程序，299

LISTSERV program, 299

加载并运行编译器，96，98 202

Load and go compilers, 96, 98 202

局域网 (LAN), 294

Local area networking (LAN), 294

洛斯阿拉莫斯[新墨西哥州]科学

Los Alamos [New Mexico] Scientific

实验室，34

Laboratory, 34

Lotus 1-2-3。请参阅Lotus Development Corporation

Lotus 1-2-3. See Lotus Development Corporation

Lotus 发展公司, 268

Lotus Development Corporation, 268

Lotus 1-2-3 软件，268–269，276，278，293

Lotus 1-2-3 software, 268–269, 276, 278, 293

LS-8 剧场控制器，135–136

LS-8 theatrical controller, 135–136

LSI。参见集成电路

LSI. See Integrated Circuits

卢布金，塞缪尔，41岁

Lubkin, Samuel, 41

Lynx，全球软件

Lynx, software for the World Wide

网络，302

Web, 302

Lyons, J. & Company, 10。另请参阅LEO

Lyons, J. & Company, 10. See also LEO

机器语言编程，200

Machine language programming, 200

MAD 编程语言，96，98，202

MAD programming language, 96, 98, 202

磁芯存储器，49–51，153，177，288

Magnetic core memory, 49–51, 153, 177, 288

发明，51

invention, 51

IBM，51，53–54，153

at IBM, 51, 53–54, 153

磁鼓存储器，38–40，69

Magnetic drum memory, 38–40, 69

磁带存储器，36，74，146，231

Magnetic tape memory, 36, 74, 146, 231

大型计算机，定义，63，75，125

Mainframe computer, definition, 63, 75, 125

曼彻斯特大学计算机，61，148，245。另请参阅Ferranti Atlas 计算机

Manchester University computers, 61, 148, 245. See also Ferranti Atlas computer

曼斯菲尔德，迈克，258

Mansfield, Mike, 258

Marchant 计算器，212

Marchant calculators, 212

Mark-8 计算机，226–227

Mark-8 computer, 226–227

马萨诸塞州蓝十字，计算，112–113

Massachusetts Blue Cross, computing, 112–113

MATH-MATIC。参见编译器

MATH-MATIC. See Compilers

Mauchly，John，13，18，24–27，85。另请参阅

Mauchly, John, 13, 18, 24–27, 85. See also

埃克特-莫奇利计算机公司

Eckert-Mauchly Computer Corporation

MAXC（多路访问施乐计算机），210，291

MAXC (Multiple Access Xerox Computer), 210, 291

马佐尔，斯坦，220–221

Mazor, Stan, 220–221

麦卡锡，约翰，154

McCarthy, John, 154

麦克唐纳·道格拉斯，211

McDonnell-Douglas, 211

麦克尤恩，多萝西，239

McEwen, Dorothy, 239

麦克法兰，哈罗德，198

McFarland, Harold, 198

世界图景的机械化，308

Mechanization of the World Picture, The, 308

医疗保险，113，169

Medicare, 113, 169

梅德林，250

MEDLIN, 250

Memorex，164

Memorex, 164

内存，计算机。请参阅RAM；

Memory, computer. See RAM;

磁芯存储器；磁性

Magnetic core memory; Magnetic

磁鼓存储器；半导体存储器

drum memory; Semiconductor memory

Mentor Graphics，281

Mentor Graphics, 281

水星计划，123

Mercury, Project, 123

Mercury Monitor，操作系统，123

Mercury Monitor, operating system, 123

梅特卡夫，罗伯特，259，291–292

Metcalfe, Robert, 259, 291–292

MICR（磁墨水字符识别）, 55

MICR (Magnetic Ink Character Recognition), 55

米克拉尔，222

MICRAL, 222

微模块，183

Micro-Module, 183

微处理器，217。另请参阅特定

Microprocessor, 217. See also specific

英特尔旗下的设备；MOS

devices under Intel; MOS

Technologies，Motorola Microprogramming，80，148–149。另请参阅

Technologies, Motorola Microprogramming, 80, 148–149. See also

IBM System/370 计算机架构，232

Architecture, computer for IBM System/370, 232

微软公司，105

Microsoft Corporation, 105

和苹果，265

and Apple, 265

创始，235–237

founding, 235–237

和 IBM，279

and IBM, 279

微软 BASIC，233–236

Microsoft BASIC, 233–236

SoftCard，265

SoftCard, 265

Windows 图形界面, 276, 285, 294

Windows graphical interface, 276, 285, 294

军事，美国，7，188。另请参阅国防

Military, U.S., 7, 188. See also Defense

部门

Department

Mill，数字设备公司总部。参见Assabet Mills

Mill, Digital Equipment Corporation headquarters. See Assabet Mills

千年虫。参见2000 年虫

Millennium bug. See Year-2000bug

米姆斯，福雷斯特，227

Mims, Forrest, 227

小型计算机，开发，124–136，191–193。另请参阅

Minicomputer, development, 124–136, 191–193. See also

建筑学，计算机定义，124，134

Architecture, computer definition, 124, 134

明尼苏达大学，299

Minnesota, University of, 299

民兵弹道导弹，181–182、187–188、221

Minuteman ballistic missile, 181–182, 187–188, 221

Hi-Rel（高可靠性）标准，181–182

Hi-Rel (high reliability) standard, 181–182

MIPS，项目，289–290

MIPS, Project, 289–290

MIT（麻省理工学院），38，129，133，140-141，158，259。另请参阅LINC；TX-0；TX-2；

MIT (Massachusetts Institute of Technology), 38, 129, 133, 140–141, 158, 259. See also LINC; TX-0; TX-2;

旋风仪器实验室，188

Whirlwind Instrumentation Laboratory, 188

林肯实验室，53，127

Lincoln Laboratory, 53, 127

项目 MAC，155，158，203–204，208

Project MAC, 155, 158, 203–204, 208

辐射实验室，141

Radiation Laboratory, 141

科技模型铁路俱乐部，129，208

Tech Model Railroad Club, 129, 208

MITS（微型仪器和遥测系统），227–231，245

MITS (Micro Instrumentation and Telemetry Systems), 227–231, 245

模式，寻址，199。另请参阅计算机架构

Modes, addressing, 199. See also Architecture, computer

莫尔斯，加尔文，104–105

Moers, Calvin, 104–105

分子电子学，181

Molecular Electronics, 181

摩尔，戈登，217

Moore, Gordon, 217

摩尔电气工程学院。参见宾夕法尼亚大学

Moore School of Electrical Engineering. See Pennsylvania, University of

摩尔定律，217，297

Moore’s Law, 217, 297

莫里斯迷你米诺，135

Morris Mini-Minor, 135

MOS（金属氧化物半导体），217，253

MOS (Metal Oxide Semiconductor), 217, 253

MOS 技术，264

MOS Technologies, 264

马赛克，软件，300–304

Mosaic, software, 300–304

摩托罗拉 68000 微处理器，275，281

Motorola 68000 microprocessor, 275, 281

Power PC 微处理器，290

Power PC microprocessor, 290

鼠标、电脑、260–261、303

Mouse, computer, 260–261, 303

MS-DOS，操作系统，269–271，285，293

MS-DOS, operating system, 269–271, 285, 293

MSI。参见集成电路

MSI. See Integrated circuits

Multics 操作系统，156–157、267、283、285

Multics operating system, 156–157, 267, 283, 285

多道程序设计，100

Multiprogramming, 100

穆尔托，比尔，277

Murto, Bill, 277

人月神话，101，169

Mythical Man-Month, The, 101, 169

NACA。参见国家航空咨询委员会

NACA. See National Advisory Committee for Aeronautics

NAS。参见国家先进系统

NAS. See National Advanced Systems

NASA（美国国家航空航天局）。另请参阅

NASA (National Aeronautics and Space Administration). See also

国家航空咨询委员会艾姆斯研究中心，113–114, 116– 118

National Advisory Committee for Aeronautics Ames Research Center, 113–114, 116– 118

计算于，122，154，159

computing at, 122, 154, 159

戈达德太空飞行中心，123，168

Goddard Space Flight Center, 123, 168

载人航天计划，123，124

Manned space program, 123, 124

马歇尔太空飞行中心，阿拉巴马州亨茨维尔，166–167

Marshall Space Flight Center, Huntsville, Alabama, 166–167

任务控制中心，123–124

Mission Control, 123–124

美国国家工程院，79

National Academy of Engineering, 79

国家先进系统（NAS），165

National Advanced Systems (NAS), 165

国家航空咨询委员会（NACA），38

National Advisory Committee for Aeronautics (NACA), 38

美国国家标准局，13，25-26，69，180

National Bureau of Standards, 13, 25–26, 69, 180

美国国家医学图书馆，250

National Library of Medicine, 250

国家物理实验室，43

National Physical Laboratory, 43

国家研究委员会，26

National Research Council, 26

美国国家科学基金会，258，295

National Science Foundation, 258, 295

国家安全局，51，65，161

National Security Agency, 51, 65, 161

美国国家半导体公司，165

National Semiconductor Corporation, 165

北约，105

NATO, 105

美国海军，140

Navy, U.S., 140

海军研究实验室，122

Naval Research Laboratory, 122

海军研究生院，223，239

Naval Postgraduate School, 223, 239

NBI，文字处理公司，255

NBI, word processing company, 255

NCR（国家现金出纳机公司），66–67，171

NCR (National Cash Register Corporation), 66–67, 171

会计机器，119

accounting machines, 119

被 AT&T 收购，171–172

acquired by AT&T, 171–172

收购 CRC，40-41

acquisition of CRC, 40–41

NCSA（国家超级计算应用中心），303

NCSA (National Center for Supercomputing Applications), 303

NEC（日本电气公司），172

NEC (Nippon Electric Company), 172

纳尔逊，西奥多，216，301

Nelson, Theodor, 216, 301

Netscape 通讯公司，303

Netscape Communications Corporation, 303

网络。参见ARPANET；计算机

Networks. See ARPANET; Computer

网络；以太网；IBM 系统网络架构

networking; Ethernet; IBM Systems Network Architecture

纽厄尔·艾伦，102

Newell, Allen, 102

纽约证券交易所，126

New York Stock Exchange, 126

NIH（国立卫生研究院），212

NIH (National Institutes of Health), 212

数码管，213

Nixie tubes, 213

尼克松，理查德·M.，121

Nixon, Richard M., 121

NLS（在线系统），301

NLS (On Line System), 301

诺布尔，大卫·L.，232

Noble, David L., 232

诺里斯·威廉，36岁，172

Norris, William, 36, 172

北美航空，2，6，187

North American Aviation, 2, 6, 187

诺斯罗普飞机公司，19–20，26

Northrop Aircraft, 19–20, 26

诺顿，约翰，235

Norton, John, 235

Novell，278，294

Novell, 278, 294

诺伊斯，罗伯特，179，184–188，198，220，258

Noyce, Robert, 179, 184–188, 198, 220, 258

纳特，罗伊，168

Nutt, Roy, 168

奥布莱恩，JA，140

O’Brien, J. A., 140

Odhner 计算器，212

Odhner calculators, 212

OEM（原始设备制造商），135–136，206

OEM (original equipment manufacturer), 135–136, 206

办公自动化的出现，255–256

Office automation, emergence, 255– 256

好利获得，172

Olivetti, 172

Programma 101 计算器，212–213

Programma 101 calculator, 212–213

奥尔森，肯尼斯·H.，127，135，140，195，255

Olsen, Kenneth H., 127, 135, 140, 195, 255

和 UNIX，247

and UNIX, 247

奥尔森，斯坦，129

Olsen, Stan, 129

OLTP（联机事务处理），251

OLTP (on line transaction processing), 251

操作系统，74，96–101。另请参阅

Operating systems, 74, 96–101. See also

CP/M；微软 Windows；

CP/M; Microsoft Windows;

MS-DOS；Multics 加载并运行，96、98、202

MS-DOS; Multics Load-and-go, 96, 98, 202

ORDVAC，18

ORDVAC, 18

組織人，141

Organization Man, The, 141

OS/MVS 操作系统, 206

OS/MVS operating system, 206

奥斯本，罗迪 F.，32 岁

Osborn, Roddy F., 32

Packard-Bell，166

Packard-Bell, 166

PageMaker，276

PageMaker, 276

乔治·帕克，258

Pake, George, 258

Palevsky，Max，165，210，249–250

Palevsky, Max, 165, 210, 249–250

帕尔默，罗伯特·B.，139

Palmer, Robert B., 139

并行处理。参见计算机体系结构

Parallel processing. See Architecture, computer

PARC。请参阅Xerox Corporation、Xerox-Parc

PARC. See Xerox Corporation, Xerox-Parc

帕克，约翰·E.，37岁

Parker, John E., 37

Pascal 编程语言，106

Pascal programming language, 106

帕特森，蒂姆，270

Paterson, Tim, 270

帕特森，戴维，288–290

Patterson, David, 288–290

PC-DOS。请参阅MS-DOS

PC-DOS. See MS-DOS

宾夕法尼亚大学，22，25

Pennsylvania, University of, 22, 25

摩尔电气工程学院, 23, 25, 45

Moore School of Electrical Engineering, 23, 25, 45

五角大楼（建筑），170

Pentagon (building), 170

计算机外围设备，153

Peripherals, for computers, 153

玻璃市，艾伦，24岁，102

Perlis, Alan, 24, 102

佩罗，H.罗斯，169

Perot, H. Ross, 169

个人电脑，293.参见Apple；

Personal computers, 293. See Apple;

Altair；IBM 个人计算机

Altair; IBM Personal Computer

Pertec 公司, 236

Pertec Corporation, 236

Philco 公司，65，111，188

Philco Corporation, 65, 111, 188

S-1000、S-2000 计算机、66、130

S-1000, S-2000 computers, 66, 130

菲尼克斯科技，277

Phoenix Technologies, 277

PL/I 编程语言，101，106–108，223，233

PL/I programming language, 101, 106–108, 223, 233

PL/M 编程语言，223，238

PL/M programming language, 223, 238

计划准备机，83–84

Plan preparation machine, 83–84

平面工艺，185–186

Planar process, 185–186

柏拉图，173–174

PLATO, 173–174

插头兼容制造商，164–165

Plug-compatible manufacturers, 164– 165

波杜斯卡，比尔，281

Poduska, Bill, 281

《大众电子》杂志，225–228，233

Popular Electronics magazine, 225–228, 233

波塞冬导弹，188

Poseidon missile, 188

POTS（“普通老式时间共享”），251

POTS (“plain old time sharing”), 251

Power PC 微处理器。参见摩托罗拉普惠公司，24

Power PC microprocessor. See Motorola Pratt & Whitney, 24

Prime，计算机公司，281

Prime, computer company, 281

印刷电路，193，195

Printed circuit, 193, 195

处理器技术，231

Processor Technology, 231

Programma 101。参见Olivetti

Programma 101. See Olivetti

编程语言、计算机。

Programming languages, computer.

另请参阅个别语言，例如，

See also individual languages, e.g.,

BASIC、Cþþ 等定义，87、94–95

BASIC, Cþþ, etc. definition, 87, 94–95

和软件，9

and software, 9

编程，术语的起源，20-21

Programming, origin of term, 20–21

近炸引信，180

Proximity Fuze, 180

PSRC（可插拔顺序继电器计算器），18

PSRC (Pluggable sequence relay calculator), 18

打孔卡 96 列，158

Punched cards 96-column, 158

计算, 16–17, 47, 77, 109

computing with, 16–17, 47, 77, 109

作为输入介质，111，119，123

as input medium, 111, 119, 123

QST杂志，224–225

QST magazine, 224–225

Quotron II，126

Quotron II, 126

辐射实验室。请参阅麻省理工学院辐射实验室

Radiation Lab. See MIT, Radiation Lab

曼哈顿下城 Radio Row 224 号

Radio Row, in lower Manhattan, 224

Radio Shack，263–264

Radio Shack, 263–264

TRS 80，型号 1个人电脑，263–264

TRS 80, Model 1personal computer, 263–264

TRS-80，型号 100

TRS-80, Model 100

笔记本电脑，279

laptop, 279

无线电电子杂志，224–226

Radio-Electronics magazine, 224–226

铁路快运代理, 228

Railway Express Agency, 228

RAM（随机存取存储器），49

RAM (random access memory), 49

动态（DRAM），198

dynamic (DRAM), 198

RAMAC。参见IBM Model 350

RAMAC. See IBM Model 350

磁盘存储单元

Disk Storage unit

Ramo-Woldridge。参见TRW

Ramo-Woldridge. See TRW

兰德公司，45，88，169

RAND Corporation, 45, 88, 169

随机存取存储器。请参阅RAM

Random access memory. See RAM

拉斯金，杰夫，273

Raskin, Jef, 273

RAYDAC 计算机，54

RAYDAC computer, 54

RCA公司，44–45，54，163，187，191，501计算机，92，162–163

RCA Corporation, 44–45, 54, 163, 187, 191, 501 computer, 92, 162–163

BIZMAC 计算机，56–57

BIZMAC Computer, 56–57

Spectra 70 系列，163

Spectra 70 series, 163

罗纳德·里根 55, 170, 289

Reagan, Ronald, 55, 170, 289

寄存器，在计算机处理器中，59–60。另请参阅

建筑学、计算机

Architecture, computer

雷明顿兰德，13，54，45。另请参阅UNIVAC（公司）Eckert-Mauchly 计算机部门，27

Remington Rand, 13, 54, 45. See also UNIVAC (Company) Eckert-Mauchly Computer Division, 27

资源一，211

Resource One, 211

RISC（精简指令集计算机），287–290

RISC (reduced instruction set computer), 287–290

里奇，丹尼斯，106，157，283

Ritchie, Dennis, 106, 157, 283

罗伯茨，H.爱德华，226–230，235–236

Roberts, H. Edward, 226–230, 235–236

罗切斯特，Nat，34岁

Rochester, Nat, 34

罗克福德研究所，104

Rockford Research Institute, 104

罗克韦尔国际，221

Rockwell International, 221

《滚石》杂志，207、210–211、258

Rolling Stone magazine, 207, 210–211, 258

ROM（只读存储器），220，240

ROM (read only memory), 220, 240

罗辛，鲍勃，96，98

Rosin, Bob, 96, 98

128 号公路，马萨诸塞州科技区，140

Route 128, technology region in Massachusetts, 140

RPG 编程语言，95

RPG programming language, 95

RSTS-11 操作系统，适用于 PDP-11、204–205、236

RSTS-11 operating system, for PDP-11, 204–205, 236

RT-11 操作系统，适用于 PDP-11，238–240

RT-11 operating system, for PDP-11, 238–240

鲁蒂豪瑟·海因茨 84 岁

Rutishauser, Heinz, 84

SABRE（半自动商业研究环境），250

SABRE (semiautomatic business research environment), 250

SAGE（半自动地面环境），51、53、67、95、127、140、154、250、260

SAGE (semiautomatic ground environment), 51, 53, 67, 95, 127, 140, 154, 250, 260

加州大学圣地亚哥分校，268

San Diego, University of California at, 268

桑迪亚国家实验室，181

Sandia National Laboratories, 181

Scantlin Electronics, Inc.，126

Scantlin Electronics, Inc., 126

Scelbi-8H 电脑，225

Scelbi-8H computer, 225

《科学美国人》杂志，225

Scientific American magazine, 225

科学数据系统 (SDS)，165，167，191，248。另请参阅Palevsky，Max

Scientific Data Systems (SDS), 165, 167, 191, 248. See also Palevsky, Max

SDS 910, 920 电脑, 166

SDS 910, 920 computers, 166

SDS 940 计算机（后来的 XDS 940）、166、210、221、246

SDS 940 computer (later XDS 940), 166, 210, 221, 246

SDS 92 型计算机，166

SDS Model 92 computer, 166

SCOOP（最优问题科学计算），31

SCOOP (scientific computation of optimum problems), 31

SEAC（标准东方自动计算机），31

SEAC (Standards Eastern Automatic Computer), 31

西雅图电脑产品，270

Seattle Computer Products, 270

Selectron 存储管，44–45

Selectron memory tube, 44–45

半导体存储器，195–198，288

Semiconductor memory, 195–198, 288

分享, 88, 100

SHARE, 88, 100

夏普，计算器，213

Sharp, calculators, 213

志摩正敏 220

Shima, Masatoshi, 220

舒加特，艾伦，232

Shugart, Alan, 232

西格尔，约瑟夫，253

Siegel, Joseph, 253

硅，由其制成的电路，183

Silicon, circuits made of, 183

Silicon Graphics 公司，172，285，290，303

Silicon Graphics, Inc., 172, 285, 290, 303

硅谷，科技区

Silicon Valley, technology region in

加利福尼亚州，140，166，258

California, 140, 166, 258

西蒙·赫伯特，102

Simon, Herbert, 102

辛普森，汤姆，124

Simpson, Tom, 124

销售点终端制造商 Singer，249

Singer, maker of point-of-sale terminals, 249

SNOBOL 编程语言, 95

SNOBOL programming language, 95

技术的社会建构，4-5

Social construction of technology, 4–5

社会保障法，113

Social Security Act, 113

工作社会学，33

Sociology of Work, The, 33

软件，术语定义，9，80-81

Software, definition of term, 9, 80–81

软件工程，105，170

Software engineering, 105, 170

固态逻辑技术。请参阅 IBM、System/360

Solid Logic Technology. See IBM, System/360

SOLO 计算机，66–67

SOLO computer, 66–67

排序，88–90，152

Sorting, 88–90, 152

SOS（SHARE操作系统），100

SOS (SHARE operating system), 100

新机器的灵魂， 246，305。

Soul of a New Machine, The, 246, 305.

另请参阅数据通用

See also Data General

太空大战，电脑游戏，207–208，210

Spacewar, computer game, 207–208, 210

SPARC 架构，290

SPARC architecture, 290

Speiser，Ambros P.，83

Speiser, Ambros P., 83

Sperry Rand，57。另请参阅雷明顿

Sperry Rand, 57. See also Remington

兰特；UNIVAC（公司）

Rand; UNIVAC (company)

斯皮尔伯格，阿诺德，56岁

Spielberg, Arnold, 56

SPOOL（同时在线外围操作），124

SPOOL (simultaneous peripheral operations on line), 124

SPREAD 委员会，146–148

SPREAD Committee, 146–148

电子表格，软件，267

Spreadsheets, software, 267

堆栈，架构，62，218

Stack, architecture, 62, 218

标准模块化系统，包装，179

Standard modular system, packaging, 179

斯坦福大学，140，207，216

Stanford University, 140, 207, 216

斯坦福人工智能实验室，207

Stanford Artificial Intelligence Laboratory, 207

斯坦福研究所，259

Stanford Research Institute, 259

斯坦福大学联网工作站，282

Stanford University networked workstations, 282

乔治·R·斯蒂比茨，70岁

Stibitz, George R., 70

存储技术, 164

Storage Technology, 164

存储程序原理，20–22。另请参阅

Stored program principle, 20–22. See also

建筑学、计算机；冯·诺依曼，约翰

Architecture, computer; von Neumann, John

结构化编程，103–104

Structured programming, 103–104

子程序，84–85，218

Subroutines, 84–85, 218

SUN 微系统公司，172

SUN Microsystems, 172

创始，281–282

founding, 281–282

策略, 286, 290, 297

strategy, 286, 290, 297

超级计算机，172。另请参阅Control Data Corporation；Cray, Seymour；Illiac-IV

Supercomputers, 172. See also Control Data Corporation; Cray, Seymour; Illiac-IV

萨瑟兰，伊万，285

Sutherland, Ivan, 285

系统开发公司（SDC），169

System Development Corporation (SDC), 169

系统网络架构。参见IBM

Systems network architecture. See IBM

Tandy Radio Shack。参见Radio Shack

Tandy Radio Shack. See Radio Shack

陶布，亨利，168

Taub, Henry, 168

税务管理系统 (TAS)，121–122。另请参阅美国国税局

Tax Administration System (TAS), 121–122. See also Internal Revenue Service

泰勒，罗伯特，259

Taylor, Robert, 259

TCP/IP（传输控制协议/互联网协议），295–296

TCP/IP (transmission control protocol/internet protocol), 295–296

科技模型铁路俱乐部。参见麻省理工学院

Tech Model Railroad Club. See MIT

技术系统，4

Technological systems, 4

Teletype 公司，227，251

Teletype Corporation, 227, 251

型号 ASR 33、124、133–134、152

Model ASR 33, 124, 133–134, 152

型号 37、251

Model 37, 251

电传，164

Telex, 164

Telnet，298

Telnet, 298

德州仪器，179，182，187–188，213，215，217

Texas Instruments, 179, 182, 187–188, 213, 215, 217

TEXT-90，IBM 7090 程序，255

TEXT-90, program for IBM 7090, 255

查克·萨克，259

Thacker, Chuck, 259

思维机器公司，300

Thinking Machines, Inc., 300

汤普森，肯，106，157，283

Thompson, Ken, 106, 157, 283

汤普森-拉莫-沃尔德里奇。参见TRW

Thompson-Ramo-Woldridge. See TRW

亨利·戴维·梭罗，312

Thoreau, Henry David, 312

分时，定义，154–155

Time sharing, definition, 154–155

TIP（终端接口处理器），297

TIP (terminal interface processor), 297

提图斯，乔纳森，225

Titus, Jonathan, 225

令牌环。参见IBM

Token ring. See IBM

TOPS-10 操作系统, 208

TOPS-10 operating system, 208

顶视图，图形界面，276

Top View, graphical interface, 276

TRAC 语言，104

TRAC language, 104

计算机中的晶体管，65，130

Transistors, in computers, 65, 130

特拉维斯·欧文 26 岁

Travis, Irwin, 26

三部曲，165

Trilogy, 165

Truoung, Thi T., 222

TRW，82，170，235

TRW, 82, 170, 235

TSO（分时选项），251

TSO (time sharing option), 251

TTL（晶体管-晶体管逻辑），192–193、217、225–226

TTL (transistor-transistor logic), 192– 193, 217, 225–226

图灵，AM，42–43，149

Turing, A. M., 42–43, 149

电视打字机，226，231

TV-typewriter, 226, 231

TX-0 计算机，56，127，243

TX-0 computer, 56, 127, 243

TX-2 计算机，262

TX-2 computer, 262

TYMNET，250

TYMNET, 250

Tymshare，166，211，259

Tymshare, 166, 211, 259

加州大学洛杉矶分校，259

UCLA, 259

英国，计算机发展，10

U.K., computer developments in, 10

安德伍德公司，41

Underwood Corporation, 41

Unisys, 171.另请参阅Burroughs；Remington Rand；UNIVAC

Unisys, 171. See also Burroughs; Remington Rand; UNIVAC

联合鞋机公司，53

United Shoe Machinery, 53

UNIVAC（公司），126，163，171。另请参阅

UNIVAC (company), 126, 163, 171. See also

雷明顿兰德；斯佩里兰德；

Remington Rand; Sperry Rand;

Unisys NTDS 计算机，126

Unisys NTDS computer, 126

固态 80 计算机，66

Solid State 80 computer, 66

UNIVAC [I] 计算机, 13, 15–16, 45, 48, 89, 217

UNIVAC [I] computer, 13, 15–16, 45, 48, 89, 217

顾客，26-28

customers, 26–28

描述，20，27–29

description, 20, 27–29

和鼓机，41–42

and drum machines, 41–42

逻辑结构, 25

logical structure, 25

内存容量，23

memory capacity, 23

打印机，31

printers, 31

使用中，30–34

in use, 30–34

UNIVAC 1103 计算机，65

UNIVAC 1103 computer, 65

UNIVAC 1108 系列, 165, 167

UNIVAC 1108 series, 165, 167

UNIVAC 文件计算机, 57

UNIVAC File computer, 57

UNIVAC II 计算机，92

UNIVAC II computer, 92

通用图灵机，149

Universal Turing Machine, 149

加州大学伯克利分校。参见伯克利

University of California at Berkeley. See Berkeley

UNIX，79，106，156，171，206

UNIX, 79, 106, 156, 171, 206

伯克利版本 (BSD)，283–284、295

Berkeley version (BSD), 283–284, 295

创造，282–283

creation, 282–283

和 DEC PDP-11，283

and DEC PDP-11, 283

和 DEC PDP-7, 283

and DEC PDP-7, 283

新手难度：260

difficulty for novice, 260

和数字设备公司，247，283-284

and Digital Equipment Corporation, 247, 283–284

和 MS-DOS，271

and MS-DOS, 271

URL（统一资源定位器），302

URL (uniform resource locator), 302

Usenet，298-299

Usenet, 298–299

用户组, 88, 215–216, 230, 264

User groups, 88, 215–216, 230, 264

苏联，计算机发展，11

U.S.S.R., computer developments in, 11

犹他大学，259，285

Utah, University of, 259, 285

乌托邦，通过计算机实现，33，237，309–311

Utopia, brought on through computers, 33, 237, 309–311

先锋计划，122

Vanguard, Project, 122

退伍军人管理局，137

Veterans Administration, 137

VHSIC（超高速集成电路），172

VHSIC (very high speed integrated circuit), 172

Viatron，252–254

Viatron, 252–254

模型 21, 253–254

model 21, 253–254

越南战争，159，258

Viet Nam War, 159, 258

Viola，网络软件，303

Viola, software for Web, 303

虚拟内存，245–246。另请参阅DEC

Virtual memory, 245–246. See also DEC

VAX；Ferranti Atlas；IBM System/370

VAX; Ferranti Atlas; IBM System/ 370

病毒，计算机，284

Virus, computer, 284

VisiCalc, 267–268, 276

VisiOn，图形界面，276

VisiOn, graphical interface, 276

Visual Basic，79。另请参阅BASIC VLSI。请参阅集成电路

Visual Basic, 79. See also BASIC VLSI. See Integrated circuits

VMS 操作系统, 246

VMS operating system, 246

冯·布劳恩，沃纳，167

von Braun, Wernher, 167

冯·诺依曼体系结构，6，23–24，42，178，216。另请参阅计算机体系结构

von Neumann architecture, 6, 23–24, 42, 178, 216. See also Architecture, computer

冯·诺依曼，约翰，21, 34, 44, 57, 59, 89

von Neumann, John, 21, 34, 44, 57, 59, 89

WAIS（广域信息服务），299–300

WAIS (Wide Area Information Service), 299–300

王实验室，212，254–257，263

Wang Laboratories, 212, 254–257, 263

LOCI 计算器，212

LOCI calculator, 212

1200型文字处理器，255

Model 1200 word processor, 255

2200 型计算计算器，255

Model 2200 computing calculator, 255

WPS 文字处理系统，256–257

WPS word processing system, 256– 257

王文字处理和个人电脑，293

Wang word processing, and personal computers, 293

王安 51, 212, 255

Wang, An, 51, 212, 255

滑铁卢大学，计算机科学系，202-203

Waterloo University, computer science dept., 202–203

WATFOR，WATFIV，WATBOL，203

WATFOR, WATFIV, WATBOL, 203

沃森，托马斯，小，44，70，288，290

Watson, Thomas, Jr., 44, 70, 288, 290

沃森·托马斯（老），67岁

Watson, Thomas, Sr., 67

波峰焊，用于组装电路板，193

Wave-soldering, for assembling circuit boards, 193

韦恩州立大学，43

Wayne State University, 43

西海岸计算机设计，40

West Coast computer design, 40

西部电气，65

Western Electric, 65

西屋电气，181，187

Westinghouse, 181, 187

惠勒跳，84

Wheeler Jump, 84

惠勒（DJ），84岁

Wheeler, D. J., 84

旋风计划，14，140

Whirlwind, Project, 14, 140

旋风电脑，59，86

Whirlwind computer, 59, 86

全球概览, 207

Whole Earth Catalog, 207

所见即所得的计算机界面，262

WYSIWYG computer interface, 262

怀特，威廉，141

Whyte, William, 141

威登哈默，詹姆斯，36岁

Wiedenhammer, James, 36

威尔克斯，莫里斯，23，148–149

Wilkes, Maurice, 23, 148–149

威廉姆斯记忆管，34–35，44–45

Williams memory tubes, 34–35, 44–45

威廉姆斯足球俱乐部，34-35

Williams, F. C., 34–35

WIMP 接口，261

WIMP interface, 261

温彻斯特磁盘存储，200

Winchester disk storage, 200

Windows，图形界面。请参阅Microsoft

Windows, graphical interface. See Microsoft

绕线，72–73，132，179

Wire-wrap, 72–73, 132, 179

木轮，43。另请参阅IBM 650

Wooden Wheel, 43. See also IBM 650

伍斯特电报和公报，电脑排版，109

Worcester Telegram and Gazette, computerized typesetting, 109

字长，计算机，58，125，151。另请参阅

Word length, computer, 58, 125, 151. See also

建筑学、计算机

Architecture, computer

小型计算机，194

minicomputer, 194

PDP-10, 208

个人电脑文字处理，268

Word processing, for personal computers, 268

万维网，300–303

World Wide Web, 300–303

史蒂文·沃兹尼亚克 264

Wozniak, Steven, 264

X-Windows，294

X-Windows, 294

Xanadu，超文本软件，301–302

Xanadu, hypertext software, 301–302

施乐公司，167

Xerox Corporation, 167

阿尔托，261–262

Alto, 261–262

8010 星信息系统, 262– 263

8010 Star Information System, 262– 263

XDS 940 计算机，222。另请参阅SDS 940

XDS 940 computer, 222. See also SDS 940

XDS 部门，167

XDS division, 167

Xerox-PARC（帕洛阿尔托研究中心），257–261，273–275，291–292，297

Xerox-PARC (Palo Alto Research Center), 257–261, 273–275, 291– 292, 297

耶鲁大学，299

Yale University, 299

威廉·耶茨，227

Yates, William, 227

2000 年错误，93，151

Year-2000 bug, 93, 151

Z4 电脑，83

Z4 Computer, 83

Zierler，N.，86

Zierler, N., 86

齐洛格，220

Zilog, 220

楚泽，康拉德，11，64，83–84

Zuse, Konrad, 11, 64, 83–84

Table of Contents

1 商业计算的出现，1945-1956 年

1 The Advent of Commercial Computing, 1945–1956

2 计算时代的到来，1956-1964

2 Computing Comes of Age, 1956–1964

3 软件的早期历史，1952-1968

3 The Early History of Software, 1952–1968

4 从大型机到小型机，1959-1969 年

4 From Mainframe to Minicomputer, 1959–1969

5. “Go-Go”年代和系统/360，1961-1975

5 The ‘‘Go-Go’’ Years and the System/360, 1961–1975

6 芯片及其影响，1965-1975年

6 The Chip and Its Impact, 1965–1975

7. 个人电脑，1972-1977年

7 The Personal Computer, 1972–1977

8. 增强人类智力，1975-1985

8 Augmenting Human Intellect, 1975–1985

9 工作站、UNIX 和网络，1981 年至 1995 年

9 Workstations, UNIX, and the Net, 1981–1995

10 “互联网时代”，1995-2001

10 “Internet Time,” 1995–2001

结论：世界图景的数字化

Conclusion: The Digitization of the World Picture